SDE @Deccan AI | BITS Pilani CS | DigiCert

Hyderabad, Telangana, India

Bengaluru, Karnataka, India

Ahmedabad, Gujarat, India




Bachelor's degree, Computer Science
Birla Institute of Technology and Science, Pilani
Join Meet and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal BossSDE @Deccan AI | BITS Pilani CS | DigiCert
I build backend infrastructure and AI systems at Deccan AI, mostly in Go and AWS. Some things I've shipped: an in-house auth service that replaced Descope and cut our auth costs by about 85%; a task orchestration service using Redis distributed locks that holds up under ~1.2K concurrent annotators; a multimodal generation pipelines (image and video) on a LiteLLM gateway that tracks spend per team and per task. Outside work I build things I want to exist. The Reading Room Orchestra (thereadingroomorchestra.club) recommends classical music for the specific book chapter you're reading. Clean cURL is a Chrome DevTools extension that copies network requests as stripped-down cURL so you can paste them into an LLM without burning tokens on junk headers. Binary-bishop is a chess engine in Python, playing at roughly 1300 Elo.. it has beaten me exactly once. BITS Pilani, B.E. Computer Science with a minor in Finance. Always up for talking about Go, LLM infrastructure, or what to listen to while reading.
Hyderabad
claude code
cursor
mcp
Fullstack Developer at Deccan AI, Dec 2024 - Present · 1 yr 10 mos, Hyderabad, Telangana, India, • Built the backend for an AI agent platform inside the annotation product (Go control-plane API, Python orchestration service, AWS Step Functions) with immutable agent versioning, schema-driven input bindings, an async run engine with bulk evaluations, and per-run token/cost breakdowns • Built a multi-modal generation layer on a LiteLLM gateway: five image/video pipelines on a submit–poll–finalize state machine (to sidestep Lambda's 15-min cap), per-workspace virtual keys, task-level spend tracking • Added fal.ai video to LiteLLM as a drop-in plugin, then upstreamed it as two merged PRs: a native fal.ai video provider and a deployment-pricing fix • Architected the auth microservice in Go (Echo) on AWS with JWT management, refresh token rotation, and RBAC, replacing Descope and cutting auth infrastructure costs ~75% • Designed the annotation task orchestration service with Redis Redlock distributed locking, supporting 1.2K concurrent users with ~20K-line JSON payloads per request • Built a GCS/S3-backed dataset ingestion pipeline mapping rowIndex to byteOffset over unstructured data, enabling random access without materializing into a DB • Designed the event tracking pipeline (SQS to workers to Brevo) handling 10K+ daily events, replacing daily batch jobs with real-time email/SMS • Developed a code execution platform on Judge0 with asynchronous test case evaluation, processing 10K+ peak submissions/day • Built an AI-powered interview platform on Retell with resume-based question generation and configurable auto-evaluation
Intern Associate Software Engineer at DigiCert, Jul 2023 - Dec 2023 · 6 mos, Bengaluru, Karnataka, India, • Parallelized Cypress E2E test automation, cutting run time by 90% (3+ hours → 30 minutes) • Fixed IDOR vulnerabilities with access control changes, securing data for 3 enterprise customers
Secretary at Gurjari Cultural Association, Sep 2022 - May 2023 · 9 mos
RPA Developer at Silver Touch Technologies Ltd, May 2022 - Jul 2022 · 3 mos, Ahmedabad, Gujarat, India
Bachelor's degree, Computer Science, Birla Institute of Technology and Science, Pilani
github: https://github.com/patel-26meet