Manishkumar Ashokkumar Patel

Rising Codementor

US$8.00

For every 15 mins

ABOUT ME

Principal Engineer with 24 years in tech and 14 years in AI/ML/DL.

Principal AI engineer with 14 years in ML/AI and 24 years in software engineering.

Available now for remote AI/ML contract and part-time engagements. Specialize in shipping LLM-powered systems end-to-end: retrieval, evaluation frameworks, prompt design, multi-provider orchestration, and production deployment.

Currently building Navay, an LLM evaluation platform that has reproduced findings from published academic research using synthetic personas.

Previously Lead Data Scientist at EPAM Systems and co-founder of Mindtree's Applied AI Research group.

English

Chennai (+05:30)

Joined May 2023

EXPERTISE

10 years experience

10 years experience

8 years experience

8 years experience

6 years experience

1 year experience

10 years experience

REVIEWS FROM CLIENTS

Manishkumar's profile has been carefully vetted and approved as a Codementor. Connect with Manishkumar now, and leave a review for them once you're done!

EMPLOYMENTS

Founder & Principal Engineer

NAVAY PRIVATE LIMITED

2025-08-01-Present

Designed, built, and shipped an end-to-end LLM evaluation platform that assembles panels of bias-corrected synthetic personas to ...

Designed, built, and shipped an end-to-end LLM evaluation platform that assembles panels of bias-corrected synthetic personas to evaluate marketing creative; reproduced findings from a published academic study (Kumar et al. 2022, Frontiers in Nutrition), matching 10 of 11 documented themes.
Implemented multi-provider LLM orchestration across OpenAI, Anthropic, and Google with cost controls, prompt caching, structured-output validation, and graceful fallback; persona library of 12,000+ synthetic personas and growing with selective panel sampling per evaluation.
Built a domain-driven FastAPI backend with MongoDB and Qdrant; Next.js + Astro frontends; deployed on Vercel with custom evaluation/observability instrumentation.
Designed an evaluation framework for LLM-generated qualitative outputs — rubric bundles, judge models, reasoning-first scoring — applicable to any GenAI product needing trustworthy quality signals.
Published reproducible validation studies (Kumar 2022 food choices; brand-creative replications for Coinbase and Hims & Hers) documenting methodology and persona pool composition end-to-end.

Python

MongoDB

System design

Python

MongoDB

System design

Design Architecture

Solo Founder

OpenAI

Fastapi

Software architecture

Google gemini

Qdrant

Anthropic

Anthropic’s model context protocol

Lead Data Scientist

EPAM Systems

2024-04-01-2025-08-01

Built an LLM-assisted incident triage PoC for high-volume scheduled data-transfer jobs: ingested logs, surfaced failure signature...

Built an LLM-assisted incident triage PoC for high-volume scheduled data-transfer jobs: ingested logs, surfaced failure signatures, and generated root-cause hypotheses with suggested fixes to reduce time-to-diagnosis for support teams.
Developed an internal NL2SQL assistant with schema grounding, query validation guardrails, and iterative evaluation with non-technical stakeholders.
Created document intelligence PoCs to extract structured fields from large PDFs, handling layout variability and confidence-based fallbacks.
Integrated anomaly detection components into internal systems with monitoring and reporting.
Line-managed 20+ engineers on the EPAM side across multiple concurrent workstreams.

Python

Google gemini

Python

Google gemini

Sr. Computer Scientist

EdgeVerve (Infosys subsidiary)

2024-01-01-2024-04-01

Led enhancement of the end-to-end ML data pipeline (data generation, model training, selection, deployment) for AI-powered enterp...

Led enhancement of the end-to-end ML data pipeline (data generation, model training, selection, deployment) for AI-powered enterprise products.

Python

Deep Learning

Langchain

Python

Deep Learning

Langchain

Llamaindex

RAG

Yolov8