Jacob Abraham

Software Engineer | 3 Years at Amazon & Kotak

Started with paper-reading sessions at MIRIx, got hooked on AI alignment. Built scalable ML solutions at Amazon, then spent time at Kotak figuring out how AI can make banking less painful. Now at Waterloo, still chasing the same questions.

View Experience See Projects
✉️ j32abrah@uwaterloo.ca 💼 LinkedIn 🔗 GitHub

Technical Skills

🔧 Backend Development

Java Spring Boot Python FastAPI REST APIs GraphQL Microservices

☁️ Cloud & Infrastructure

AWS Lambda DynamoDB S3 Athena EKS Docker

🤖 Machine Learning & AI

TensorFlow PyTorch Qiskit Neural Architecture Search Quantum ML

📊 Data & Observability

SQL Data Pipelines Prometheus Grafana ETL

What I Bring to Your Team

🚀 From Contests to Production: Cut my teeth on competitive programming and hackathons, then applied that problem-solving mindset to build systems serving millions at Amazon and Kotak.
🔬 Research + Shipping: Final year project on reward side-channels in RL. Published AI safety work with MIRIx. But also shipped real features—UPI Circle, automated loan pipelines, data infrastructure.
🎓 Leadership & Community: IEEE Computer Society Chair as a sophomore (our chapter was recognized as world's best in 2020). Google DSC Campus Lead. Grew IEEE membership 28% through AI workshops.

Featured Projects

Highlighted work demonstrating backend engineering and ML research capabilities

Backend + ML

🧠 Friends Circle Platform

Full-stack platform using vector embeddings for intelligent friend matching. Built with production-grade backend architecture.

  • FastAPI backend with <500ms query latency
  • QDrant vector database for semantic search
  • GraphQL API for efficient data fetching
  • AI-powered event scheduling recommendations
FastAPI Python QDrant GraphQL NextJS
View Project →
ML Research

🔐 Quantum ML for Network Security

Research comparing quantum ML methods (QCNN, QNN, VQC, QSVM) against classical baselines for intrusion detection.

  • Implemented 5 quantum ML architectures
  • Benchmarked vs. SVM, Decision Tree, Random Forest
  • Demonstrated improved feature representation
Quantum ML TensorFlow Qiskit Python
View Repository →
ML Research

⚙️ Zero-Cost Neural Architecture Search

Novel approach replacing training-based accuracy with zero-cost proxy scores in genetic algorithm optimization.

  • Modified NSGA-Net framework
  • Eliminated need for model training during search
  • Reduced architecture search time significantly
NAS Genetic Algorithms TensorFlow Python
View Work →
AI Safety

🔬 Coherent Extrapolated Volition

Group study with MIRIx on CEV and its flaws. We posted our analysis in a respected AI safety community, sparking discussion on the implications and potential fixes.

AI Safety Research MIRIx
Learn More →
Full Stack

🧠 Mentem - Mental Health Platform

Healthcare platform with intelligent matching and personalized recommendations for mental health support.

Python Healthcare ML
View Repository →
Game Theory

🎮 Aumann Agreement Game

Born from a MIRIx paper discussion—a rationality calibration game where players converge to agreement through Bayesian updating. Built it to make the theorem tangible.

JavaScript Game Theory
View Repository →

Work Experience

AWS

Software Development Engineer

June 2022 – April 2023

Amazon Web Services - Bangalore, India

  • Automated data transformation pipelines — designed Athena-based catalog generation, eliminating 60% of manual overhead for data teams
  • Built distributed job allocation system — developed tooling for managing workloads across fully managed AWS infrastructure with real-time metadata insights
  • Optimized supply chain operations — shipped feature reducing manual steps by 30%, directly impacting Amazon's operational efficiency
Stack: Python, AWS (Lambda, Athena, S3, DynamoDB), CDK, Data Pipelines
Kotak

Software Development Engineer

May 2023 – August 2025

Kotak Mahindra Bank - Bangalore, India

  • Built UPI Circle service from scratch — designed delegated payment flows ensuring transaction consistency across millions of users in India's largest payment network
  • Reduced loan processing from 2-3 days → 3 hours — automated E-Loan pipeline with intelligent data workflows, cutting manual review effort by 70%
  • Improved system observability by 40% — integrated Prometheus/Grafana monitoring into backend services, dramatically reducing mean-time-to-diagnosis
Stack: Java Spring Boot, AWS (EKS, S3, DynamoDB), Athena, Docker, Prometheus, Grafana

Education

M.Eng

Computer Science & Engineering

University of Waterloo

Waterloo, ON, Canada

Sept 2025 – Dec 2026

Focus: Artificial Intelligence & Machine Learning

B.Tech

Computer Science & Engineering

College of Engineering Trivandrum

Trivandrum, India

Aug 2018 – May 2022

Get In Touch

Let's connect and discuss how I can contribute to your project or team.

Email GitHub LinkedIn