About
Karan Prasad
I'm a developer and entrepreneur with deep experience in cloud and DevOps. I focus on building dependable backend systems, shipping automation that keeps teams moving, and turning messy problems into useful products. After years as a backend engineer, I'm now shifting more of my energy into hands-on machine learning.
As the founder of Obvix Labs (obvix.io), an AI safety and governance studio delivering guardrail proxies, red-team rituals, and grounded agents for regulated teams, I partner with startups and founders to simplify their tech journey. We build custom SaaS products, robust cloud infrastructures, and intelligent automation tools designed to help you launch and grow responsibly.
My approach is shaped by a multidisciplinary background in technology, design, and business. If you're an investor, founder, or innovator passionate about AI, automation, or system design, let’s connect.
Portfolio
Obvix Lake
AI-native customer intelligence lake that stitches knowledge graphs, multichannel sentiment telemetry, and self-learning ticket loops to deflect 70–80% of routine inquiries, cut support cost up to 30%, and reach ROI inside 12 months.
Obvix Proxy
Drop-in LLM guardrail proxy that intercepts prompts, replies, and tool calls with rule + AI reasoning, external lookups, and containment so teams get ~58 ms latency safety shields plus audit-ready telemetry.
Obvix Red
Adversarial red-teaming lab running prompt-injection gauntlets, persona-vector attacks, agentic tool-abuse drills, and governance evidence lockers aligned to EU AI Act, NIST AI RMF, and DPDP controls.
Full‑Stack Conversational AI Engine
Built a self-hosted, persona-driven conversational AI stack using transformers, RAG, and RandomForest to boost conversions, personalize journeys, and optimize CTAs through real-time sentiment/emotion analysis and adaptive ML techniques.
Bodhi AI
AI assistant that converts static FAQs into dynamic, real-time conversations to ease support workloads.
NeevTrace
DLT-powered platform delivering transparency and sustainability intelligence for Indian supply chains. Awarded the MLH MongoDB track.
End-to-End Ethereum Holder Scraper
Built an end-to-end pipeline for scraping ethereum token holders and getting their socials. Used Puppeteer,Face detecion,and some cleaver System-Design.