Darshil Shah
Portfolio / 2026Data Scientist · Machine Learning · GenAI & LLMs

I build LLM and machine-learning systems that make it to production.

(01)OverviewIn brief

Data scientist building production machine-learning and GenAI systems since 2019: agentic LLM pipelines, retrieval and fine-tuned language models, and the predictive models behind payments and lending, along with the evaluation, tracing and deployment that keep them honest.

Currently
Data Scientist II, Aviso AI
Previously
Visa, PayU, MINRO Lab (IIIT Bangalore), Jekson Vision
Education
MS in IT (Data Science), IIIT Bangalore
Core stack
Python, SQL, PyTorch, LangGraph, LangChain, RAG, FastAPI, AWS
(02)Capabilities6 disciplines
  • 01Languages

    PythonSQL

  • 02Machine Learning

    Scikit-learnCatBoostPandasPySparkClassificationRankingPropensity ModellingLead Scoring

  • 03Deep Learning

    PyTorchTensorFlowKerasHuggingFaceTransformersBERTSentence-BERTPEFT / LoRANERComputer VisionOpenCVCUDA

  • 04Generative AI & LLMs

    LangGraphLangChainRAGAgentic PipelinesFew-shotPrompt EngineeringLLM-as-a-JudgeLLM Fine-tuningOpenAIGeminiClaudeAmazon BedrockVoice Agents

  • 05Cloud & MLOps

    AWSEC2S3ECRTrueFoundryDockerGitCI/CDLLM Observability & Tracing

  • 06Backend & Data

    FastAPIPostgreSQLpgvectorAstraDBVector Search

(03)Projects3 live

Built end to end. Live to try.

LLM Pipeline · Healthcare · Full-stack
Live

The last consultation, in ninety seconds.

A web app for doctors: record or upload a consultation and get back a structured clinical summary, a care plan and a follow-up reminder, ready before the patient's next visit.

  • Record in the browser or upload audio; speech-to-text with speaker labels turns it into a doctor / patient transcript
  • A three-step LLM chain extracts what was said, analyses it, then writes a brief the doctor can read in 90 seconds
  • “Come back in three weeks” becomes a dated follow-up in a queue, which closes itself when the next visit is booked
  • A complete product: FastAPI backend, Next.js frontend, Postgres, Google sign-in, and each doctor sees only their own patients
  • Python
  • FastAPI
  • LangChain
  • Gemini / Bedrock
  • Deepgram
  • Postgres (Neon)
  • Next.js
Try it liveGoogle sign-in
10 more from Aviso AI, Visa, PayU and researchSee all projects→
(04)Experience2019 — Present
  • —Context-Aware Summarization: designed and built an 11-node LangGraph agentic pipeline that classifies calls across Sales, Post-Sale, Internal and Outbound, then routes each call to LLM workflows for context-aware summaries and insights.
  • —Built LLM-driven participant role assignment, call-type / sub-type classification and domain-specific insight generation, bringing CRM, participant and revenue-intelligence context into summarization.
  • —Deployed via TrueFoundry with AWS ECR and LLM tracing for production observability.
  • —Smart Trackers: built a human-in-the-loop, few-shot LLM framework that lets clients create custom conversation trackers from initial phrases, retrieving semantically similar historical sentences from AstraDB and annotating them Yes / No into tracker-specific few-shot datasets.
  • Python
  • LangGraph
  • LangChain
  • LLMs
  • AstraDB
  • AWS
  • TrueFoundry
  • LLM Observability
(05)Publications4 peer-reviewed
(06)Education
Aug 2020 – Jan 2023
IIIT Bangalore logo

IIIT Bangalore

Master of Science in IT

Data Science, Machine Learning, Deep Learning · GPA 3.78

International Institute of Information Technology · Bangalore, India

Jun 2015 – Jun 2019
Nirma University logo

Nirma University

B.Tech in Instrumentation & Control

Institute of Technology, Nirma University · Ahmedabad, India

(07)ContactBangalore, India