
Anmol Vijay Bhatia
I build AI agents that ship to production
I'm a guy with a head full of schemes and a heart full of memes. My obsession with tech started with absurd hours in Minecraft, which quickly spiraled into cracking open hardware and teaching myself system modding to see how things actually worked under the hood. To me, computer science is the ultimate universal tool, sitting right at the intersection of systems, art, human behavior, and problem-solving. After earning my bachelor's in CS and spending three years engineering software in the field, I'm now pursuing my Master's in AI to double down on the technology I'm most passionate about. I'm actively seeking full-time roles and internships where I can tackle complex problems, build intelligent systems, and help move the needle forward.
Technical Skills
Work Experience
Software Engineer (AI/ML Enablement)
Incedo Inc.
This role turned AI/ML from a passion into my daily focus. I engineered the ATOLL Network Summarizer, a multi-agent LangGraph pipeline that shrank telecom data analysis from hours to seconds, and built a multi-agent NL-to-SQL system that translates natural language into self-validated database queries. To keep these systems safe and predictable, I shipped a custom MCP server to strictly sandbox agent execution and audit logging. Alongside hands-on development, I partnered directly with the Chief Solutions Officer to define the long-term AI product strategy.
Associate Software Engineer (Full Stack)
Incedo Inc.
Before shifting to AI, I spent over two years learning what enterprise scale really means while supporting a platform of 150,000+ users. My focus was on performance, security, and system reliability. I unblocked redundant API dependencies to plummet page load times from 40 seconds down to 2 seconds, rebuilt our Role-Based Access Control (RBAC) system from scratch, cutting unauthorized access incidents by 85%, and modernized a rigid legacy UI into customizable, user-driven dashboard widgets. I also engineered a batch-digest pipeline for transactional emails, eliminating rate-limiting bottlenecks and server strain.
Computer Science Teaching Assistant
University of Illinois at Chicago - College of Engineering
I led labs for over 130 computer science students. Instead of passive lecturing, I anchored sessions around pair programming and Test-Driven Development (TDD) to build real-world coding habits early. My biggest emphasis was systematic debugging, coaching students to step back and isolate root causes methodically rather than guessing, whether they were untangling recursive logic, troubleshooting pointer arithmetic, or stress-testing custom data structures.
Mathematics Learning Assistant
University of Illinois at Chicago - Department of Mathematics
In my math and theory sections, I noticed students often mechanically followed formal proofs without truly grasping why they worked. To fix this, I flipped my approach, building intuitive, conceptual understanding first before introducing formal notation, which completely transformed their confidence. I also maintained a weekly feedback loop with lead instructors, tracking recurring student pain points to continually adapt and refine our upcoming sessions around what actually needed clarification.
Intern
KPMG Kuwait
My engineering background started with systems auditing and infrastructure health. I used tools like Burp Suite and Wireshark alongside custom scripts to hunt down web vulnerabilities and patch them proactively. I also evaluated older IT systems, leveraging Gartner market research to make a data-driven business case for executive modernization priorities.
Featured Projects

Deep Agentic Research Engine
Give it a question and it does the research for you — generating sub-queries, crawling sources, checking them against each other, and writing up a synthesized report. It routes across OpenAI, Gemini, and Anthropic depending on the task and streams everything back over SSE, so you watch it think instead of staring at a spinner.
Multi-step agentic loop: query generation, crawling, verification, synthesis
Multi-provider routing across OpenAI, Gemini, and Anthropic
Streamed, source-grounded reports via SSE

Legal Clause Auditor LLM
I wanted to know if a small, self-hosted model could do real legal work without paying an API per call — so I fine-tuned LLaMA 3 8B with QLoRA on California court rulings to detect legal clauses, hitting 89% accuracy. I took it the whole way: domain adaptation, GGUF quantization, and a FastAPI service that runs with zero dependency on any external LLM provider.
QLoRA fine-tuning of LLaMA 3 8B on California court rulings
89% accuracy on legal clause detection
Quantized to GGUF and self-hosted via FastAPI

Document Intelligence System
Legal and financial forms are where most OCR pipelines fall apart — layout matters as much as text. I built a multi-modal pipeline combining LayoutLMv3 and Donut to read the structure and the content together, pulling out key fields at 90% accuracy — 30% better than the traditional OCR-plus-NLP approach on the same documents.
LayoutLMv3 + Donut for layout analysis and key-info extraction
90% extraction accuracy, 30% better than OCR + NLP
Applied to real legal and financial forms

Ecommerce Inventory Management System
A full-stack inventory platform I built end-to-end for 5 stakeholders — schema and backend from scratch, wrapped in a dashboard non-technical users could actually operate. It tightened up their supply-chain pipeline and cut spend that was leaking out of a manual process.
Engineered a scalable MySQL schema with Spring Boot + Hibernate ORM
Real-time updates and an analytics dashboard
Deployed on AWS EC2/RDS with Docker

ReliaNet
I rebuilt reliable networking from scratch on top of lossy UDP — stop-and-wait ARQ, sequence-numbered ACKs, EWMA-based RTT estimation — basically to understand everything TCP quietly does for you, by doing it myself. Then I hammered it across 50+ scenarios with configurable packet loss and latency to prove it held up.
Stop-and-wait ARQ with sequence-numbered ACKs
EWMA-based RTT estimation for adaptive timeouts
Validated across 50+ configurable loss/latency scenarios

15 Puzzle AI Solver
The project that made search algorithms click for me. I implemented IDA* to solve the classic 15-puzzle, then pitted two heuristics against each other — Manhattan distance vs. misplaced tiles — across hundreds of configurations to see where each one wins. Every run prints the moves, nodes expanded, time, and memory, so the theory becomes something you can actually watch.
IDA* search with Manhattan distance and misplaced-tile heuristics
Benchmarked admissibility vs. efficiency across hundreds of configurations
Reports move sequence, nodes expanded, time, and memory usage
Education
Master of Science in Artificial Intelligence (Incoming)
New Jersey Institute of Technology
September 2026 - December 2027
Bachelor of Science in Computer Science, Minor in Mathematics
University of Illinois at Chicago
August 2019 - May 2023
Awards & Certifications
R&R Program - Individual Excellence Award
Incedo Inc. • October 2025
Recognized for consistent high-impact contributions to the Digital Engineering team across client-facing AI initiatives and enterprise platform delivery.
R&R Program - Individual Excellence Award
Incedo Inc. • June 2024
Selected from the Digital Engineering division for exceptional individual performance and for raising the bar.
"You inspire us everyday!" - Nitin Seth, CEO Incedo Inc.
AWS Certified Cloud Practitioner
Amazon Web Services Training and Certification • May 2025
Demonstrated cloud fluency and foundational AWS knowledge. Able to identify essential AWS services necessary to set up AWS-focused projects.
Volunteering & Leadership
Mentor
SparkHacks - UIC Annual Hackathon
March 2023
Mentored student teams on architecture tradeoffs and project scoping, helping reframe vague ideas into buildable systems with clear evaluation criteria within a 24-hour timeline.
Team Lead
Code Day
February 2020
Taught problem decomposition over syntax to beginners, helping students build and ship a working project within a single day through real-time pair debugging and structured iterative feedback.
Orientation Leader
Shorelight (University of Illinois Chicago)
August 2022 - September 2022
Developed a comprehensive orientation week schedule for all incoming students. Conducted engaging seminars on University policies, with tailored advice for CS students to excel in their academics.
Volunteer
Autism Aashram • Hyderabad, India
Helped care for children and supported their daily activities.