Ownership,
not tickets.

About the role

Vamstar is looking for an Entry Data Analyst to join our growing global team. This is an excellent opportunity for recent graduates or early-career professionals to build hands-on experience in data research, analytics, and healthcare markets, while working on projects that have real-world, global impact. The role is ideal for candidates who are detail-oriented, analytical, and eager to learn in a fast-paced, collaborative environment.

What you will own

• Perform accurate data capture from diverse data sources for the AI Agents.
• Research and analyse new markets, trends, and datasets.
• Interpret and validate large volumes of data with high accuracy.
• Maintain and improve existing data repositories and processes.
• Work extensively with Excel/Google Sheets (advanced formulas) and analytics reports.
• Deliver high-quality outputs within defined timelines.
• Collaborate with global teams to support process improvements and knowledge sharing.
• Bring attention to detail to quality checking, analysing and evaluating AI-generated outputs.

Required skills & qualifications

• Bachelor’s degree in Economics, Statistics, Pharma, or Mathematics (preferred).
• Strong analytical, research, and problem-solving skills.
• Proficiency in Google Sheets, MS Excel, and MS Word.
• Close attention to detail and numerical accuracy.
• Basic understanding of data management and analytics concepts.
• Ability to manage deadlines and work independently in a remote setup.
• Positive attitude with a willingness to learn and adapt.
• Exposure to the healthcare, pharmaceuticals, or medical devices industries.
• Interest in healthcare markets and global supply chains.
• Fresh graduates or candidates with less than 2 years of experience are encouraged to apply.

What we offer

• Two fixed weekend days off.
• Working hours: 10:00 AM – 7:00 PM IST.
• Fully remote working.
• Group health insurance.
• CTC 4 LPA.

Role overview

Own the complete ML lifecycle for an enterprise LLM system that selects and executes browser-based tools. You will turn raw execution logs into high-quality training data, run fine-tuning experiments, build rigorous evaluations, and deploy and optimise models on AWS. This is an applied ML engineering role, not prompt engineering, analytics or model-ops-only work.

ML data and evaluation

• Build pipelines to ingest, parse, clean, normalise, enrich and deduplicate execution logs and documents.
• Convert raw logs into structured supervised examples with context, candidate tools, expected actions and reference labels.
• Create versioned training, validation and holdout datasets with privacy, schema, label and quality checks.
• Design leakage-aware splits using task- or session-level boundaries.
• Build reproducible evaluations for tool-selection accuracy, reference accuracy, hallucinations, disagreements and failure slices.
• Trace misleading metrics back to inputs, labels, dataset versions and model versions.

Model development

• Benchmark open-weight models such as Qwen, Llama and Mistral on quality, training effort, long-context behaviour and inference cost.
• Run supervised fine-tuning, knowledge distillation and parameter-efficient fine-tuning experiments.
• Apply sound experiment design: clear baselines, controlled comparisons, reproducible configurations and error analysis.
• Identify overfitting, leakage, sampling bias, label noise and uneven task coverage.
• Maintain versioned datasets, prompts, training configurations, checkpoints and evaluation code.
• Recommend which experiments justify additional training or inference cost.

ML platform and MLOps

• Take models through deployment, testing, monitoring and controlled release.
• Benchmark serving configurations for throughput, latency, GPU memory, concurrency and cost at an agreed quality level.
• Evaluate vLLM or comparable runtimes, INT8 quantisation, checkpoint compatibility and KV-cache requirements.
• Measure cost per successful task or action, not just cost per token.
• Implement model versioning, CI/CD, shadow inference, disagreement logging, feature-flag releases and rollback.
• Monitor data drift, quality regressions, serving errors, latency and cost.

LLM workflow engineering

• Build pipelines involving document parsing, chunking, embeddings, vector indexing and retrieval.
• Implement function or tool calling, multi-step workflows, retries, fallbacks and state handling.
• Design observability across prompts, retrieved context, model outputs, tool calls and downstream failures.
• Write production-grade, tested Python for large-scale document and execution-data processing.

Essential qualifications

• 7+ years of Python experience.
• 5+ years of AWS experience.
• 7+ years across ML, data, backend or applied AI engineering, with production delivery experience.
• Strong supervised ML fundamentals, including validation design, overfitting, leakage, sampling bias and error analysis.
• Hands-on transformer fine-tuning using PyTorch, Hugging Face or equivalent tooling.
• Direct ownership of a completed model training, fine-tuning, distillation or benchmarking project.
• Experience preparing ML/LLM datasets and building model evaluation code.
• Experience with experiment tracking, model versioning, CI/CD, deployment, monitoring and rollback.
• Working knowledge of GPU inference, quantisation, batching, concurrency, tail latency and cost optimisation.
• Experience with large-scale data processing using Spark, Kafka, Flink or comparable technologies.
• Strong SQL and practical experience with relational and NoSQL systems.
• Production experience with RAG, embeddings, vector databases, function calling or agentic workflows.
• Ability to work autonomously and own outcomes without detailed specifications.

Preferred background

• Startup or similarly high-ownership environment.
• AWS S3, Glue, IAM, EC2 GPU instances, CloudWatch and model registries.
• vLLM, NVIDIA GPU optimisation or AWS G5-family instances.
• Qwen, Llama or Mistral model families.
• LoRA, QLoRA, knowledge distillation or long-context evaluation.
• Browser automation, test execution logs or tool-selection workflows.
• Enterprise or regulated environments where privacy, auditability and governance matter.

What to include with your application

• A model you trained, fine-tuned, distilled or benchmarked, including the dataset, validation approach, metrics and your contribution.
• An ML data, evaluation, deployment or inference system you owned, including scale, failure handling and measurable results.
• Your current Indian city, earliest start date and interview availability.

Core responsibilities & objectives

• Design, build, and maintain batch/streaming data pipelines: ingestion, cleaning, normalisation, enrichment, deduplication.
• Build and own ML/LLM pipelines end-to-end: document and log parsing, chunking, embeddings generation, vector indexing, agentic tool calling, multi-step workflows, retries, fallbacks, and state handling.
• Turn raw execution logs and documents into reliable, versioned training and evaluation datasets, with leakage-aware splits and measurable quality gates.
• Build reproducible evaluation harnesses: measure tool-selection accuracy, hallucination rates, and failure slices, not just aggregate metrics.
• Write production-grade, well-tested Python that processes large volumes of data and documents reliably.
• Own pipeline health: if data is stale, broken, or wrong, it’s on you.
• Work autonomously to project deadlines with minimal hand-holding.

Key qualifications & skills (non-negotiable)

• 7+ years in backend data-heavy development or data engineering.
• Previously worked in a startup.
• Highly proficient in Python, with 7+ years of work experience.
• 5+ years of work experience with Amazon Web Services (AWS).
• Hands-on experience with large datasets and high-velocity data streams (Kafka, Flink, Spark).
• Strong with pipeline orchestration tools (Airflow, MLflow, or equivalent).
• Solid SQL skills (Postgres, BigQuery, or Snowflake) and NoSQL experience (DynamoDB, OpenSearch, Elastic).
• Real experience with LLM workflows: RAG architectures, embeddings/vector DBs, prompt engineering, function/tool calling, observability.
• Hands-on LLM dataset preparation and evaluation: deduplication, sampling, train/validation splitting without task or session leakage, and tracing bad metrics back to inputs and labels.
• Applied experience with supervised fine-tuning or model benchmarking, and an understanding of quality-vs-cost trade-offs.
• Deep understanding of ETL/ELT patterns and data processing at scale.

Preferred background (strong signals)

• Experience with the AWS data stack at scale (S3, Glue, EC2/GPU instances).
• Exposure to enterprise or regulated environments where data governance matters.
• Built and shipped data, ML and LLM-powered pipelines in production.
• Experience with knowledge distillation, parameter-efficient fine-tuning (LoRA/QLoRA), or long-context evaluation.
• Familiarity with model serving concerns (vLLM, quantisation, latency/cost trade-offs), even if you haven’t owned inference infrastructure.
• Has debugged a pipeline and knows why observability matters.
• Worked in a fast-moving startup where “that’s not my job” doesn’t exist.

What will get you rejected

• An “I set up the pipeline, someone else monitors it” mindset.
• Tutorials and side projects but no production experience at scale.
• Prompt-only AI experience: you’ve never built or evaluated the data and training pipeline behind the model.
• Can’t explain trade-offs between streaming vs. batch, why you chose one vector DB over another, or how you split training data to avoid leakage.
• Needs detailed specs before writing a line of code.
• No curiosity about what the data actually means or how the model uses it.

The role

• Gather client business requirements to build and maintain end-to-end data reporting pipelines and visual dashboards.
• Maintain and improve our existing tools and processes with statistical methods and machine learning.
• Apply data management and business intelligence methodologies.
• Work to project deadlines efficiently.
• Lead project deliverables and manage direct client communications.
• Collaborate closely with the Senior Data Engineer to build and maintain AWS-based data pipelines.
• Work comfortably in an ambiguous, fast-moving startup environment with evolving processes.
• Bridge technical pipeline work with client-facing reporting and insights.

Required skills

• Advanced knowledge of Excel, Power BI, Tableau.
• B.Tech or technical degree with 4+ years of experience, including 4+ years of hands-on AWS.
• SQL and Python expertise, with 3+ years of work experience in Python.
• 2+ years of work experience with Amazon Redshift.
• Writing reusable code and preparing automation for future analyses and QA.
• Advanced prompt engineering and LLM coding.
• High level of numerical accuracy and comfort working with large data sets.
• Prior experience with finance and pricing analytics.
• An interest or experience in healthcare and medical sectors with a focus on pricing is a strong plus.
• Proficiency with Git/GitHub for version control and collaborative development.
• Data validation, reconciliation, and audit-trail practices for high-stakes data.

The details

• Location: UK only (right to work required; no sponsorship). Remote-first, 2–3 days a month in person, plus periodic travel to customer sites in Europe (France, Poland, Italy, the Nordics).
• Type: contract, 6 months initial with extension expected and an option to convert to permanent. We’d rather convert than re-recruit.
• 5 days a week preferred; 4 considered for an exceptional candidate.

Core responsibilities & objectives

• Own the requirements practice, not just the requirements. Set the standards for discovery, user stories, acceptance criteria, and Definition of Ready across all Polaris enterprise programmes, and enforce them, including sending back work raised by people more senior than you.
• Lead and grow the BA function. Mentor business analysts on the programmes, review their output, and build the reusable assets (discovery templates, story patterns, acceptance script templates, a requirements playbook) that let the practice scale.
• Own the “what and whether it’s right” on the most complex programmes. The Delivery Manager owns the when. You own requirement validation end-to-end: running discovery with customer tender, pricing, sales and commercial teams; testing every requirement against the mapped process, platform capability, and what the work order committed to; and saying so, in writing, early where they diverge.
• Separate what the customer asked for from what the customer needs. Propose the Polaris behaviour that meets the need, get the customer to sign the requirement, not the wish, and protect scope. Requests outside the work order are flagged as changes, with sized impact, before anyone agrees to them.
• Master the product to the depth of its behaviour: how discovery sources are enabled per market, how matching is tuned and what its confidence means, how assessment criteria are configured, how exports and integrations are shaped.
• Own acceptance quality. UAT scripts and pass/fail definitions written before the build starts; the verdict on whether a delivered feature meets the criteria sits with your function.
• Make value measurable. Every programme’s pilot KPIs defined with an agreed baseline and measurement method, so the final pilot report is written from data, not anecdote.
• Feed the product. Turn what programmes learn into structured feedback: what is customer-specific, what should become platform, what keeps coming up.

Key qualifications & skills (non-negotiable)

• 12+ years as a business analyst on enterprise software, with at least 5 on B2B SaaS or data products delivered to large corporate customers. Not internal IT change, not consultancy bench rotation.
• Has led, mentored or line-managed other business analysts, and can point to a requirements practice you built or materially raised the bar on.
• Has owned requirements end-to-end for at least two enterprise implementations in parallel, each with its own contract, stakeholders and milestones.
• Has told a senior customer stakeholder that what they asked for was not what they needed, proposed the alternative, and got it signed.
• Your stories do not generate calls: you can show a redacted story an engineer built from without a conversation and a tester verified without interpretation.
• Reads contracts: work orders, statements of work, scope exhibits, milestone-based billing.
• Data literate: confident with Excel, comfortable in SQL or equivalent, able to define a KPI and sanity-check the number that comes back.
• Works with distributed engineering teams across time zones; your writing is good enough to be the primary channel.
• Holds ground calmly with fast-moving leadership and customers with budgets and deadlines. Direct, early, with options.
• UK-based with right to work in the UK, able to travel in Europe.

Preferred background (strong signals)

• Public procurement or tendering, especially in MedTech or pharma: lots, framework agreements, technical dossiers, award notices.
• Requirements for AI/LLM-based products: defining “good enough” for a probabilistic output, confidence thresholds and fallbacks.
• Integration requirements for CRM, ERP or tender-management systems (Dynamics 365, SAP, Salesforce, or sector-specific tools).
• Working proficiency in French, Polish, Italian, German or a Nordic language.
• Experience in a scale-up where you built the requirements practice rather than inheriting one.
• Prior contract-to-permanent conversions, with a reference from the company that converted you.

Core responsibilities & objectives

• Design, build, and maintain enterprise-grade integrations end-to-end: connecting Microsoft Dynamics 365, Salesforce, and the wider commercial tech stack into Vamstar’s agentic AI platform.
• Architect integration patterns across CRM, CPQ, pricing, and contracting ecosystems (APIs, event-driven messaging, ETL/ELT pipelines, and middleware), ensuring reliable, scalable, and secure data flows.
• Build and own connectors and data pipelines for Salesforce, SharePoint, Snowflake, Vendavo, Databricks, and Contract Lifecycle Management (CLM) systems, with proper error handling, retries, idempotency, and reconciliation.
• Implement and operate integrations on AWS (Lambda, Step Functions, S3, EventBridge/SQS, API Gateway, Amazon AppFlow, or equivalent), with strong observability across the entire integration layer.
• Own integration health: if a sync fails, a schema drifts, or data doesn’t reconcile, it’s on you to detect, diagnose, and fix it.
• Work autonomously to project deadlines with minimal hand-holding, collaborating directly with client IT teams and Vamstar’s product and engineering groups.
• A few onsite client visits.

Key qualifications & skills (non-negotiable)

• 7+ years in integration architecture/engineering, shipping production-grade enterprise integrations.
• Previously worked in a startup or fast-moving consulting environment.
• Deep, hands-on expertise with Microsoft Dynamics 365 (Dataverse, Web API, Power Platform connectors, or equivalent) and Salesforce (REST/Bulk APIs, platform events, Apex, MuleSoft or equivalent).
• Strong data engineering fundamentals: pipelines with Snowflake and Databricks, including schema design, incremental loads, and data quality checks.
• Hands-on experience integrating with SharePoint (document libraries, Microsoft Graph API) and CLM platforms (e.g. DocuSign CLM, Icertis, Ironclad, Agiloft).
• Production experience on AWS at scale, including serverless and event-driven architectures; Amazon AppFlow is a strong plus.
• Strong programming skills in Python, Node.js/TypeScript, or equivalent, with solid knowledge of REST, webhooks, OAuth/authentication patterns, and message queues.
• Deep understanding of integration observability: logging, monitoring, alerting, and end-to-end tracing across systems.

Preferred background (strong signals)

• Experience with Vendavo or other enterprise pricing/CPQ platforms: pricing data models, quote-to-cash flows, margin analytics.
• Exposure to healthcare, life sciences, or regulated industries (e.g. GDPR, HIPAA-aware development, validated systems).
• Familiarity with integration platforms (MuleSoft, Boomi, Workato, Azure Logic Apps, or equivalent), and when not to use them.
• Has debugged a production integration incident across CRM, data warehouse, and middleware layers.
• Comfortable with DevOps basics: Docker, infrastructure-as-code (Terraform/CDK), CI/CD pipelines, monitoring and alerting.
• Expertise in supply chain, RFP, tender, marketplaces, web-scale B2C/B2B.

Core responsibilities & objectives

• Design, build, and migrate Ab Initio workflows to AWS Glue/PySpark.
• Write production-grade, well-tested Python that processes large volumes of data and documents reliably.
• Use AI coding tools (e.g. Claude) hands-on to generate PySpark / AWS Glue code and build test harnesses.
• Own pipeline health: if data is stale, broken, or wrong, it’s on you.
• Work autonomously to project deadlines with minimal hand-holding.

Key qualifications & skills (non-negotiable)

• 7+ years in backend data-heavy development or data engineering.
• Highly proficient in Python, with 5+ years of work experience.
• 3+ years of work experience with Amazon Web Services (AWS).
• 2+ years of work experience with Ab Initio, including production deployments and cloud migrations.
• Hands-on experience with large datasets and high-velocity data streams (Kafka, Flink, Spark).
• Strong with pipeline orchestration tools (Airflow, MLflow, or equivalent).
• Solid SQL skills (Postgres, Redshift, or Snowflake) and NoSQL experience (DynamoDB, OpenSearch, Elastic).
• Real experience with LLM workflows: RAG architectures, embeddings/vector DBs, prompt engineering, function/tool calling, observability.
• Deep understanding of ETL/ELT patterns and data processing at scale.

Preferred background (strong signals)

• Experience with the AWS data stack at scale.
• Exposure to financial services industries.
• Built and shipped data, ML and LLM-powered pipelines in production.
• Has debugged a pipeline and knows why observability matters.
• Worked in a fast-moving startup where “that’s not my job” doesn’t exist.

Core responsibilities & objectives

• Design, build, and maintain full-stack applications end-to-end: responsive frontends, RESTful/GraphQL APIs, backend services, and data layers.
• Build and own user-facing features that surface complex AI/LLM capabilities: document viewers, search interfaces, agentic workflow dashboards, real-time streaming responses, and multi-step workflow UIs with proper error states, retries, and fallback handling.
• Write production-grade, well-tested code across the stack (TypeScript/React frontend, Python/Node.js backend) that handles large volumes of data and documents reliably.
• Own feature health: if a page is slow, an API is broken, or a user flow is wrong, it’s on you.
• Work autonomously to project deadlines with minimal hand-holding.

Key qualifications & skills (non-negotiable)

• 7+ years in full-stack development, shipping production web applications.
• Previously worked in a startup.
• Highly proficient in modern frontend frameworks (React, Next.js, or equivalent) and TypeScript.
• Strong backend development experience (Node.js) building scalable APIs and services.
• Hands-on experience with databases and data-heavy applications (Postgres, Redis, Neptune or equivalent) and designing efficient queries and schemas.
• Real experience integrating AI/LLM capabilities into products: streaming LLM responses, RAG-powered search UIs, function/tool calling, async job handling, and observability.
• Deep understanding of application architecture, CI/CD, testing strategies, and performance optimisation at scale.

Preferred background (strong signals)

• Experience with AWS at scale (Lambda, ECS, S3, CloudFront, or equivalent).
• Exposure to healthcare, life sciences, or regulated industries (e.g. GDPR, HIPAA-aware development).
• Built and shipped AI/LLM-powered features in production products used by real customers.
• Has debugged a production incident across frontend, API, and data layers, and knows why observability matters.
• Comfortable with DevOps basics: Docker, infrastructure-as-code, monitoring and alerting.
• Worked in a fast-moving startup where “that’s not my job” doesn’t exist.
• Expertise in supply chain, RFP, tender, marketplaces, web-scale B2C/B2B.

Le poste

• Lieu : France uniquement, télétravail, temps plein.
• Rattachement hiérarchique : Directeur de l’Ingénierie.
• Clients existants : entreprises du Top 100 des sciences de la vie, des technologies médicales et de l’industrie pharmaceutique.
• Type de poste : Freelance/CDD, temps plein. Langues : anglais niveau professionnel.

Principales responsabilités et objectifs

• Concevoir, développer et maintenir des pipelines de données par lots (batch) et en flux continu (streaming) : ingestion, nettoyage, normalisation, enrichissement et déduplication.
• Concevoir et gérer de bout en bout des pipelines ML/LLM : extraction et parsing de documents, segmentation, génération d’embeddings, indexation vectorielle, appels d’outils, workflows multi-étapes, nouvelles tentatives, mécanismes de repli (fallback) et gestion de l’état.
• Développer du code Python robuste et testé, capable de traiter de grands volumes de données et de documents de manière fiable.
• Garantir la qualité des pipelines : la détection et la correction des données obsolètes, corrompues ou erronées relèvent de votre responsabilité.
• Travailler en autonomie et respecter les délais des projets avec un minimum d’encadrement.

Qualifications et compétences clés (non négociables)

• Plus de 7 ans d’expérience en développement backend ou en ingénierie des données.
• Expérience préalable en startup.
• Excellente maîtrise de Python, React et Node.
• Expérience pratique des grands ensembles de données et des flux à haute vélocité (Kafka, Flink, Spark).
• Maîtrise des outils d’orchestration de pipelines (Airflow, MLflow ou équivalent).
• Solides compétences en SQL (PostgreSQL, BigQuery ou Snowflake) et expérience NoSQL (DynamoDB, OpenSearch, Elastic).
• Expérience concrète des workflows LLM : architectures RAG, embeddings et bases vectorielles, ingénierie des prompts, function calling, observabilité.
• Compréhension approfondie des modèles ETL/ELT et du traitement des données à grande échelle.

Profil souhaité (fortement pertinent)

• Expérience de la pile de données AWS à grande échelle.
• Connaissance des secteurs de la santé, des sciences de la vie ou des industries réglementées.
• Expérience de la conception et de la mise en production de pipelines de données, de ML et de LLM.
• A déjà débogué un pipeline et comprend l’importance de l’observabilité.
• A travaillé dans une startup dynamique où « ce n’est pas mon travail » n’a pas sa place.

Core responsibilities & objectives

• Design, build, and maintain batch/streaming data pipelines: ingestion, cleaning, normalisation, enrichment, deduplication.
• Build and own ML/LLM pipelines end-to-end: document parsing, chunking, embeddings generation, vector indexing, agentic tool calling, multi-step workflows, retries, fallbacks, and state handling.
• Write production-grade, well-tested Python that processes large volumes of data and documents reliably.
• Own pipeline health: if data is stale, broken, or wrong, it’s on you.
• Work autonomously to project deadlines with minimal hand-holding.

Key qualifications & skills (non-negotiable)

• 7+ years in backend data-heavy development or data engineering.
• Previously worked in a startup.
• 3+ years of work experience with Amazon Web Services (AWS).
• Highly proficient in Python, with 5+ years of work experience.
• Hands-on experience with large datasets and high-velocity data streams (Kafka, Flink, Spark).
• Strong with pipeline orchestration tools (Airflow, MLflow, or equivalent).
• Solid SQL skills (Postgres, BigQuery, or Snowflake) and NoSQL experience (DynamoDB, OpenSearch, Elastic).
• Real experience with LLM workflows: RAG architectures, embeddings/vector DBs, prompt engineering, function/tool calling, observability.
• Deep understanding of ETL/ELT patterns and data processing at scale.

Preferred background (strong signals)

• Experience with the AWS data stack at scale.
• Exposure to healthcare, life sciences, or regulated industries.
• Built and shipped data, ML and LLM-powered pipelines in production.
• Has debugged a pipeline and knows why observability matters.
• Worked in a fast-moving startup where “that’s not my job” doesn’t exist.

What will get you rejected

• An “I set up the pipeline, someone else monitors it” mindset.
• Tutorials and side projects but no production experience at scale.
• Can’t explain trade-offs between streaming vs. batch, or why you chose one vector DB over another.
• Needs detailed specs before writing a line of code.
• No curiosity about healthcare or what the data actually means.