Hi, my name is
K V Shanmukha Sai.
Architecting robust backend solutions.
Software engineer specializing in scalable microservices, high-performance APIs and agentic AI systems, and one half of The Hot Path on YouTube.
Software engineer specializing in scalable microservices, high-performance APIs and agentic AI systems.
Currently — Software Engineer, National Center for Ontological Research

01 — About
A laid-back, results-driven engineer.
I have a Master’s in Computer Science from SUNY Buffalo. I design scalable microservices, optimize high-performance REST APIs, and build intelligent systems with agentic frameworks. I like owning complex problems and mentoring teammates, and I bring React skills when the front end needs it.
Outside tech: stock markets, current politics, social issues, and photography.
Frameworks
- Node.js
- Express.js
- NestJS
- Next.js
- FastAPI
Languages
- JavaScript
- TypeScript
- Java
- Python
Databases
- MySQL
- MongoDB
- Redis
- Elasticsearch
- PostgreSQL
Infra & tools
- Kafka
- Docker
- AKS
- Azure Service Bus
- GitHub Actions
02 — Experience
Where I’ve worked
Mar 2026 — Present
Software Engineer
National Center for Ontological Research
Core developer of Ontos Cosmos, an applied-AI platform that turns any input — streaming data, audio, PDF, Excel, RDF/Turtle — into governed, fully traceable structured knowledge; generated 1,000+ knowledge graphs across 9,400+ entities, cutting expert modeling work from ~1 month to ~15 minutes.
May 2025 — Mar 2026
AI Backend Engineer
Built the transcription service behind a product used by 5,000+ users across author and academic communities — converting audio and video to text at 95% accuracy with speaker attribution and sub-2-minute turnaround, plus LLM post-processing for transcript cleanup and summaries.
Jul 2022 — Jul 2024
Software Development Engineer I & II
Built the personalization rule engine that decided what content and health messaging each user saw across the homepage and wellness pages, serving 300+ enterprise clients, and made patient-record search ~60% faster across 10M+ records with CQRS and Elasticsearch.
Jan 2022 — Jun 2022
Software Development Engineer Intern
Cut content-update dependency from 15 days to 2 by delivering a CMS portal with React and a headless CMS.
Mar 2021 — May 2021
Full-stack Intern
Designed and shipped a MERN website for the client The Investico in 10 days, deployed on AWS.
Dec 2020 — Mar 2021
Software Development Intern
Built the cart-management backend for mobile covers and merchandise with Node.js and MongoDB.
03 — Videos
On YouTube
I co-create The Hot Path: system design, explained in depth. These are my latest videos.
Why AI Types One Word At A Time | Agentic AI 101 #13
Run AI Offline — On Your Own Laptop | Agentic AI 101 #12
MCP, Explained — I Let AI Apply to Jobs For Me
Shorts
quick ones, under a minute
Langfuse: trace every LLM call to find the step that broke
Stop eyeballing prompts — test them side by side with promptfoo
Transcribe audio locally for free with faster-whisper — no API, no upload
Fine-tune an LLM on a FREE Colab GPU with Unsloth
Stop hand-tuning prompts — let DSPy optimize them for you
04 — Projects
Things I’ve built
An API mocking and management platform: mock, validate and forward requests, with JWT auth, versioning, request logging and per-project analytics.
A microservice e-commerce backend on Node.js with MongoDB, Elasticsearch, Kafka and Redis, plus LLM tooling that automates unit tests and API docs.
PySpark ETL pipelines that cluster 130K+ urban tree records into geospatial planting recommendations using K-Means, regression and MapReduce.
A distributed C++ TCP chat system with offline message buffering, user blocking, broadcast messaging and real-time routing using select().
Predicts machine anomalies from engine and sensor data using Principal Component Analysis and autoencoders. Published as a chapter in a Springer book.
A platform that helps people find doctors and symptom information, and predicts diseases with 95% accuracy.
Compared 9 machine learning algorithms for detecting phishing websites. Published on SSRN.
A web portal where sponsors explore startup ideas and back ventures, built to scale past 10,000 products.
An online voting platform with face-recognition identity checks, which raised participation by over 30%.
Research comparing lip-biometric algorithms and their compatibility with mobile operating systems. Published in IRJET.
06 — Education
Learning pathways
Master’s, Computer Science & Engineering
Bachelor’s, Computer Science & Engineering