About
Who I am
I'mfinishingmyMaster'sinBigDataAnalyticsatVITVellore,butmostofwhatIknowaboutbuildingAIsystemscamefromshippingthem,notfromasyllabus—chasingthegapbetweenamodelthatworksinademoandonethatsurvivescontactwithrealusers.
Most of my work sits at the intersection of LLMs and infrastructure: fine-tuning transformers with QLoRA, building agentic pipelines in LangGraph, and shipping them as FastAPI services with the evaluation and observability (RAGAS, LangSmith, drift detection) to know when they're actually working.
I care most about the parts most people skip: grounding, failure recovery, and honest evaluation. My agentic systems retry and reformulate instead of failing silently, and I've first-authored research on making models small and fast enough to run on real hardware, not just benchmarks.
Focus
What I work with
- 01LLM Agents
- 02RAG
- 03Fine-Tuning (QLoRA)
- 04LangGraph
- 05LangChain
- 06LlamaIndex
- 07CrewAI
- 08OpenAI API
- 09MCP
Experience
Where I've worked
- 01
AI/ML Engineer Intern
PGAGI
Aug 2026 - Present
Engineering core pieces of an agentic AI/ML platform: building model-routing and orchestration logic across LLMs and specialized agents, wiring up vector- and graph-based retrieval for document understanding and recommendations, and standing up the cloud infrastructure and security controls that keep it running in production.
- 02
Machine Learning Intern
Tensaw Technologies
Sept 2025 - April 2026
Fine-tuned a T5 transformer to 95.7% accuracy for insurance payer-plan classification, deployed a sub-second FastAPI inference API on AWS EC2 serving production traffic, and built a LangGraph-based voice assistant that cut operator handling time by 35%.
Education
Where I studied
- 01
Master's in Big Data Analytics
VIT Vellore
2024 - 2026
- 02
Bachelor's in Computer Science Engineering
St. Peter's Institute of Higher Education and Research
2020 - 2024
Currently Building
- 01
Agentic AI systems
The ideas behind Veritas-Agent: multi-agent pipelines with grounding, validation, and failure recovery.
- 02
Efficient, edge-ready ML
Small, fast architectures that run on real hardware, not just benchmarks.
- 03
Production LLM infrastructure
RAG retrieval, evaluation, and observability for systems that need to be trustworthy, not just impressive.
Stack
Tools I keep in the loop
Let's buildsomethingintelligent.
Have a project, a role, or just an idea worth kicking around? I'm always up for a conversation about agentic systems, ML infra, or anything in between.
Based in
Hosur, India
Local time
--:--IST