AI/ML Engineer · intelligent systems
Building reliable AI systems that move from prototype to production.
I work on agents, retrieval, memory, and multimodal systems — with a focus on the engineering details that make intelligent products useful and dependable.
A practical point of view
The interesting part is rarely the model alone. It's the system around it: retrieval, tools, evaluation, recovery, and the people who rely on the result.
Based in India · remote-friendly · open to meaningful AI/ML work.
Tools I use to make systems dependable
Proof of work
Outcomes worth looking at twice.
67%
agent-memory latency reduced
12s to 4s through parallelized query rephrasers
8,000+
memory cases evaluated
accuracy and retrieval behavior validated across test cases
40%
deployment latency reduced
production-minded improvements in the DevOps Agent
Python · LangGraph · LangSmith
01DevOps Agent - Autonomous Deployment Assistant
An LLM-powered DevOps agent that automates application deployments and configuration management across AWS, GCP, and Docker. Built with LangGraph and LangSmith, with Pinecone and Neo4j supporting vector retrieval and graph reasoning. Unit testing, defect analysis, and performance optimisation improved deployment reliability and reduced latency by 40%.
Read the case studyNext.js · Redis · PostgreSQL
02DeepDSA - AI Technical Interview & Assessment Platform
A microservices-based AI interview platform designed to scale to millions of concurrent users. The platform combines speech-to-text, LLM-powered natural language technical interviews, code evaluation, and skill scoring to improve assessment efficiency by 65%.
Read the case studyPython · CNNs · DNNs
03Amazon ML Challenge - Product Price Prediction
An ensemble product price prediction model combining CNNs, DNNs, and natural language processing. Sentence Transformers, XGBoost, and multimodal feature fusion were used to predict prices from product descriptions and images, achieving an SMAPE score of 56 and a top-percentile ranking.
Read the case studySynopsys Inc.
Research & Development Intern
2026-01-01 — Present
Reduced memory-step latency by 67% (12s to 4s) by parallelising three query rephrasers with conditional result merging. Validated agent-memory accuracy across approximately 8,000 test cases by integrating graph-based memory retrieval and self-contained query rephrasing. Improved multi-step agent reliability with dependency-aware sequential execution for code generation and execution workflows, and built a corrective query tool for targeted recovery. Collaborated with development and testing teams to integrate new Ansys Mechanical APIs, resolve test issues, and integrate Copilot capabilities into the Scripting Agent.
About
I like the space between an idea and the moment it becomes dependable.
I started with the visible part of software — products people could use — and kept moving toward the harder questions underneath: how systems reason, how they recover, and how we know when they are working.
Today that means building AI systems with enough structure to be useful in the real world: clear interfaces, measurable behavior, and engineering choices that hold up beyond the demo.
Capabilities
The disciplines behind the result.
Intelligent systems
Reliability and evaluation
Applied ML
Production engineering
Current toolkit: Analytical thinking · Creativity · Git · GitHub · MySQL · PostgreSQL · Docker · AWS
Writing will appear here once the first post is published.
A good next conversation