Shreyas Prasad

Senior Software Engineer at IBM, building distributed systems where infrastructure meets AI.

About

I work on IBM's engagement with the U.S. Postal Service, building the pipelines, data platforms and models behind trucking operations and contract cost analysis. More and more of that work sits where infrastructure meets AI: putting LLMs into production pipelines, and building the durable, observable systems they need to run reliably. If you're building in that space, I'd like to hear from you.

Selected work at IBM

60 min → <2 minBatch redesign onto Kafka and a worker pool on OpenShift, on 50K-record runs.
65% → 98%Field accuracy of an LLM parsing layer for scanned contracts, 5,000 documents a month.
1,500+ contractsCosted with a vehicle allocation model that became the standard for cost analysis.
$5MM+ a yearIn billing discrepancies surfaced by truck-lease invoice reconciliation.

Projects

Aug–Sep 2026

agentd

A control plane that runs LLM agents as durable, resumable, sandboxed and observable jobs.

Go
Postgres
pgvector
Docker
OpenTelemetry
MCP

Write-up Repo

Jun–Jul 2026

Foreflow

Event-driven automation for prediction markets, matching live Polymarket and Kalshi data against trigger workflows in real time.

Go
NATS
Kubernetes
Postgres
Prometheus
Next.js

Write-up Repo

Apps

Solfa Echogame
An ear-training game. Hear a run of solfège notes, tap it back, and add one more note each round.
Anagramsgame
A daily word game. Make as many words as you can from seven letters in 60 seconds.

Background

MS Computer Science: Georgia Tech · Dec 2025 · 3.8 GPA
BS Computer Science and Applied Mathematics & Statistics: Stony Brook University · May 2023

Languages: Go, Python, TypeScript, Java, C++, SQL
Infrastructure: Kafka, OpenShift, Kubernetes, Docker, PostgreSQL, NATS, Prometheus, OpenTelemetry
AI: LLM extraction, hybrid retrieval (pgvector + BM25), agent runtimes, MCP, evals

Contact

Email is best: contact@shreyasprasad.dev. I'm also on GitHub and LinkedIn.