Software Engineer & AI Systems Builder
Vedant Misra_
MS CSE @ Penn State '26. Building AI agents, distributed systems, and the infrastructure that makes them reliable at scale. Previously at Recursion Pharmaceuticals, ZS Associates, SAS Institute & EY.
Software Engineer — AI Systems & Distributed Infrastructure
Vedant Misra
I build AI systems that work in production — LLM orchestration pipelines, graph-backed retrieval engines, chaos engineering frameworks. Currently finishing an MS in Computer Science at Penn State (May 2026), with prior industry experience at Recursion Pharmaceuticals, ZS Associates, and SAS Institute. I care deeply about the intersection of distributed systems and applied AI: not just models that work in notebooks, but systems that hold up under load, failure, and real-world complexity.
MY TECHNICAL LEXICON
Skills
A systems map of the languages, frameworks, AI tooling, cloud platforms, and data technologies I use to build production-minded software. Hover or focus a node to trace its links; click to pin details.
Central profile
Vedant Misra
Software Engineer
My academic journey
Education
My educational journey has provided me with a robust foundation in computer science and software engineering.
The Pennsylvania State University
2024 to 2026
Master of Science in Computer Science & Engineering. Focus areas: AI/ML systems, computer vision, distributed computing, security, and software engineering at scale.
My professional journey
Experience
$ cat experience.log
SOFTWARE ENGINEERING INTERN
Recursion Pharmaceuticals
> 2025
Built self-serve lab operations tooling and an MCP server for AI-assisted workflows within Recursion's TrekSeq sequencing metadata ecosystem — reducing operational toil for scientists and Ops teams at one of the world's largest TechBio companies.
What I've built
Featured Projects
systemsCompleted coursework projectBuilding a Thread-Safe Channel Implementation in C
A deep dive into concurrency and synchronization
A Go-inspired channel system in C that enables safe message passing between concurrent producers and consumers, including blocking, non-blocking, close, and select-style operations.
5000+
Test Iterations
mlActive prototype / MVP-like systemCogniLink
Workspace Intelligence Platform - Hybrid RAG, Knowledge Graphs, and Operational Intelligence
A full-stack workspace intelligence platform that ingests engineering data from tools such as GitHub, Slack, Jira, Google Drive, Gmail, Calendar, and Notion into a Neo4j graph and Pinecone vector index for cited, relationship-aware AI answers.
Neo4j
Graph Memory
mlCompleted (Hackathon MVP)Northrop Grumman Object Detection
Multi-Model Ensemble Object Detection Hub
A comprehensive web-based object detection system developed for the 2024 Northrop Grumman Hackathon, integrating YOLOv8, YOLOv7, and OWL-ViT for zero-shot localization.
3
AI Models
My thoughts and ideas
Latest Blog Posts
Research
Adapting the Segment Anything Model Family for Specialized Domains
A comprehensive review of how Meta's Segment Anything Model (SAM) adapts to remote sensing, biological imaging, and video segmentation.
Read More →Get in touch