Headline

building next-gen fashion-tech | maxing out my attention heads | ai engineer

About

The childhood magic of making a computer print "Hello, World!" never really left me. It just evolved. These days, the "magic" looks a lot like spinning up backend systems that don't buckle under 10,000+ concurrent users, engineering RAG pipelines so fast they feel like cheating, and taming massive LLMs to solve problems they weren't explicitly trained for. My happy place is where robust backend engineering (think Python, Go, FastAPI) meets the wild frontier of AI . I'm a firm believer that elegant code is only half the story; it's the people and the problems that give it purpose. When I'm not maxing out my attention heads (both the model's and my own), you can find me watching or playing chess ♟️ or on the never-ending quest for the perfect pizza. Let's talk shop. Hit me up if you want to geek out about distributed systems, the future of AI agents, or why my next chess opening should be the Sicilian Defence.

Location

New Delhi

AI stack

antigravity

codex

Work experience

Senior Product Engineer at shoppin', Oct 2025 - Present · 11 mos, Gurugram, Haryana, India

AI Engineer at Sidecar, Jul 2025 - Oct 2025 · 4 mos, California, United States, - built ai agents for logistics to quote and reach out to freight forwarder agents

Founding AI Engineer at VerbaFlo.AI, Dec 2024 - Sep 2025 · 10 mos, Gurugram, Haryana, India, - built and scaled a multi-modal agentic conversational ai platform (chat, email, whatsapp, voice) handling 20k+ concurrent sessions - optimized hybrid rag pipelines with rerankers and disk ann, enabling high concurrency with sub-10ms retrieval and lower infra cost - shipped production-grade llm infrastructure with modular prompts, jinja templating, eval + observability, and fault-tolerant fallbacks

Full Stack Engineer (Contractual) at Mostly Events, Sep 2024 - Dec 2024 · 4 mos, Remote, - built the entire job platform backend from scratch, owning system design and core product features - developed a feature-rich lms with quizzes, assignments and secure video streaming using hls (m3u8)

SDE 1 at Convin, Aug 2024 - Nov 2024 · 4 mos, Bengaluru, Karnataka, India, - built scalable data pipelines enabling real-time analytics on large production datasets using pinot and kafka, debezium, golang - shipped backend features in django to improve system performance and enterprise reliability

Education

Bachelor of Technology - BTech, Computer Science, Sushant University

High School Diploma, Computer Science, Laxman Public School

Public links

github: https://github.com/harshkedia177