Ritik Verma

I'm a Backend Engineer

About

Welcome! I'm Ritik Verma, a graduate student with a strong background in Computer Science, currently pursuing an M.S. at the University at Buffalo. I have demonstrated abilities in creating state-of-the-art ML model architectures, developing database scanners, and collaborating with cross-functional teams to deliver impactful software solutions.

My journey blends strong academic grounding with hands-on professional engineering experience in backend development and full-stack projects using a variety of modern programming languages and technologies.

Backend Engineer & ML Researcher

  • Birthday: 17 April 1999
  • Phone: +1 (631) 366-9859
  • Email: ritikver@buffalo.edu
  • Age: 26
  • Degree: Master of Science (M.S.)

Resume

Education

M.S. in Computer Science

Jan 2025 - Dec 2026

University at Buffalo (SUNY), Buffalo, New York

Coursework: Algorithms, Machine Learning, Computer Vision, Deep Learning, Distributed Systems, Data Models.

B.Tech in Information Technology

July 2018 - June 2022

Techno Main Salt Lake, Kolkata, India

Coursework: Data structures, Algorithm analysis, DBMS, Operating Systems, Computer Network, Compiler Design.

Technical Skills

  • Languages: Java, Go, Python, JavaScript
  • Frameworks: Maven, Mockito, JUnit, Agile (Scrum)
  • Technologies: Github, AWS, React, Docker, Kubernetes
  • Operating Systems: Windows, Linux (Ubuntu)
  • Databases: MySQL, Cloudera, Redshift, Teradata, Snowflake

Professional Experience

Graduate Teaching & Research Assistant

May 2025 - Present

University at Buffalo

  • Teaching Assistant: TA for CSE-474 Intro to Machine Learning.
  • AIEmoCare: Architected a state-funded production platform using React (TypeScript), FastAPI, and TimescaleDB for time-series storage; deployed via Kubernetes and Docker.
  • Real-Time Fusion Engine: Engineered a low-latency (200ms) ingestion pipeline using WebSockets, Redis Pub/Sub, AsyncIO, and NVIDIA Triton to orchestrate parallel GPU inference.

Backend Engineer (Contract)

July 2024 - November 2024

Mercor

  • Achieved a 10% accuracy improvement on internal LLM benchmarks via automated data curation and hyperparameter optimization.
  • Streamlined development workflows by containerizing environments.

Full Stack Developer Trainee

Feb 2024 - April 2024

Codelogicx

  • Developed responsive e-commerce dashboards using React.js/Node.js and integrated RESTful APIs to reduce page load latency.

Associate Software Engineer

July 2022 - December 2023

Informatica

  • Improved Redshift database scanner extraction efficiency by over 40% and added Materialized view support.
  • Built a Cloudera (Hive CDP) Scanner from scratch, adding unit/integration tests and backend support.
View Full Resume

Academic Projects

Here are some of my recent technical architectures and research frameworks.

BrainDiffNet: Generative AI Framework for EEG-to-Image

Python, PyTorch, Docker (Oct 2025)

Engineered a production-ready, modular CLI framework decoupled into three stages. Reduced trainable parameters by 90% and minimized GPU memory overhead by implementing FP16 precision. Integrated MMDiT and Temporal Masked Autoencoders for high-dimensional time-series data.

AtomSigNet: Biosignal Diffusion with Markovian Reasoning

Python, PyTorch, TCN (Jan 2026)

Filed a US Patent for a Wearable Biosignal System. Developed a Markovian Reasoning Engine to autonomously identify "logic gaps" in sensor data, triggering conditional diffusion for artifact correction. Achieved 91% SOTA accuracy on WESAD/CASE datasets.

AFairDNet: Fair Multisensor Emotion Recognition System

Python, PyTorch, Hugging Face (Aug 2025)

Engineered a "Chain-of-Thought" control loop evaluating synthetic signal fairness using CLIP, dynamically refactoring LLM prompts to correct bias. Co-developed the complete framework (accepted at BSN 2025), outperforming SOTA baselines by 4% F1-score.

SynCoT: Synthetic Data Pipeline

Python, PyTorch, Gemini API (Sep 2025)

Architected a state machine coordinating a Generator, Evaluator, and Reasoner to autonomously correct data bias. Built a robust text-to-signal pipeline utilizing BERT tokenization and a custom 1D UNet, integrated with the Google Gemini API to parse fairness scores.

Contact

Mobile:

+1 (631) 366-9859