AI/ML Engineer · Massachusetts, USA
Building predictive systems, recommendation engines, and scalable ML products that move from notebook to production.
A journey in five coordinates
Origin · Kathmandu
Born in Kathmandu, where curiosity, movement, and ambition first shaped how I looked at systems and problem solving.
Bachelor’s · VIT, Tamil Nadu
Built my Computer Science foundation and learned how software, data, and systems solve real-world problems.
Machine Learning Engineer · Vivma Software
Worked on recommendation systems, dynamic pricing, preprocessing, and production ML workflows tied to business outcomes.
MS Data Analytics · Research Assistant
At Clark University in Worcester, I deepened analytics while contributing as a Research Assistant, connecting data, research, and applied AI problem solving.
AI/ML Engineer · Humana
Building healthcare AI/ML systems across claims, risk assessment, predictive analytics, and scalable deployment workflows.
Professional experience
My work spans predictive analytics, recommendations, real-time inference, and MLOps. The common thread is measurable value, resilient systems, and thoughtful delivery.
Humana
Vivma Software Inc.
Capstone Intelligence Lab
Multimodal AI capstone combining vision, action recognition, and LLM reasoning.
Built a hybrid pipeline that samples video frames, detects objects, recognizes actions, generates candidate captions, and compares LLM outputs using evaluation metrics.
Final description
Selected projects
NLP · Embeddings · RLHF
Resume parsing and candidate matching with fine-tuned SentenceTransformer embeddings, plus an adaptive technical interview practice system.
LightGBM · Neural Networks
A multi-class system predicting attack types across imbalanced threat categories, optimized through repeated stratified cross-validation.
Cutler Capital Management · Research automation
From market documents to investment-ready research packets.
This Agentic AI-style research automation model transformed daily market documents, filings, PDFs, research reports, and market materials into structured briefs, ticker priorities, summaries, and analyst-ready outputs. Final research remained human-reviewed to support investment judgment.
Collect PDFs, filings, market research, and daily source documents.
Pull company, market, REIT, convertible, and macro signals.
Connect evidence across documents and identify investment relevance.
Prioritize tickers, themes, risks, and portfolio-relevant developments.
Generate structured daily research briefs and dashboard summaries.
Keep final outputs human-reviewed for investment judgment.
> INGEST source documents queued · provenance retained
Working stack