Revanth Darshan D R

I am an AI and ML engineer at Tata Consultancy Services, focused on agentic systems and applied machine learning that hold up in production. I build LLM pipelines and process automation, along with the evaluation, guardrails and deployment that make them reliable.

Current role

Systems Engineer at Tata Consultancy Services, working as an AI Process Orchestrator: end-to-end business process automation across document generation, database operations and UI layers, with RPA extracting requirements from legacy systems that have no API.

Agentic AI

Built a Gen AI agentic requirements generator that extracts and structures business requirements from meeting transcripts and documents. It advanced to the regional level of the TCS internal AI hackathon.

Shipped

Reman, a Rust CLI that makes shell history searchable by meaning for developers and AI agents, with a built-in MCP server and at most 5 ms of overhead per command. GreenCode AI at TCS, an LLM-based tool that flags energy-heavy code in VS Code and at commit time through pre-commit quality gates.

Research

LFAT, a lightweight deepfake detector combining spatial and frequency-domain features: 99.34% accuracy on Celeb-DF with 5.6M parameters, over four times smaller than the ResNet baseline it outperformed. GEM, a memory layer for personal AI that updates dependent facts when one changes: 100% on 37 derived facts, 27 points ahead of flat vector memory.

In production

A community platform built and run end to end, live with 30+ members in English, Telugu and Tamil: React Native, FastAPI, PostgreSQL, Redis, Docker and a Play Store release pipeline. ClaimCompanion, a multi-agent insurance claims assistant where deterministic rules decide every verdict and the model only explains it.

Download résumé (PDF, 407 KB) · open in browser