Hi, I'm Saikiran Reddy Jakka
Software Engineer
MS Computer Science graduate student at Stony Brook University. Interested in working with scalable backend systems, advanced machine learning, and building innovative AI-driven applications.
About.
Background
I am a graduate researcher and software engineer specializing in distributed systems and applied machine learning. Based at Stony Brook University, my academic and professional work is driven by a focus on building systems that are theoretically sound, performant, and reliable under scale.
Currently, I serve as a Research Assistant developing supply chain intelligence platforms driven by Ergo AI. I work on bridging the gap between statistical machine learning and constraint-based symbolic reasoning, ensuring that AI-driven policy decisions remain verifiable and logically consistent.
My engineering ethos revolves around understanding systems from first principles—whether that entails implementing consensus algorithms from raw specifications or architecting cloud-native data pipelines with strict consistency bounds.
Education
Master of Science in Computer Science
Coursework: Distributed Systems, Machine Learning, Operating Systems, Advanced Algorithms.
Bachelor of Technology in Information Technology
Graduated with Honors. Focused on Database Management and Core System Design.
Research Focus
Distributed Systems & Consensus
Investigating fault-tolerant architectures and highly available state machines. Deeply interested in the performance characteristics and safety proofs of Paxos, PBFT, and Raft derivatives in adversarial network conditions.
Machine Learning & Verifiable AI
Focusing on the intersection of deep learning and symbolic logic to create explainable, constrained AI pipelines. Researching robust time-series forecasting and inventory optimization for complex supply chains.
Scalable Infrastructure Design
Studying the orchestration of distributed deployments using GitOps and infrastructure-as-code principles. Focused on zero-downtime microservices and cloud-native resilience.
Technical Stack.
Building intelligent retrieval systems and multi-agent architectures.
Architecting fault-tolerant, high-throughput database and consensus systems.
Automating resilient, observable deployments across cloud providers.
Choosing the right tool for the job across systems, backend, and ML domains.
Designing real-time streaming pipelines and high-performance storage systems.
Building production-grade backends and responsive frontend interfaces.
Automating build, test, and deploy pipelines for reproducible releases.
Implementing authentication, authorization, and audit systems from scratch.
Machine Learning & AI
My ML work started with NLP and recommendation pipelines at HCL Technologies, where I built sentiment classifiers and feature extraction workflows using scikit-learn and PyTorch. At Stony Brook, I took it further designing Graph-Tree RAG, a hybrid retrieval pipeline combining FAISS vector search, BM25, and graph traversal that improved multi-hop accuracy by 37%. I also built a Self-Evolving Agentic AI system with semantic memory and multi-LLM orchestration, eliminating 60% of manual workflow steps through adaptive prompt strategies.
Experience.
Career Path
Building scalable infrastructure and intelligent systems across academia and enterprise.
Graduate Research And Teaching Assistant
Software Development Intern
Data Analyst Intern
Get the highlights.
A deeper dive into the projects that define my technical journey.
Ergo AI: Intelligent Supply Chain Optimization
Building a supply chain intelligence platform powered by Ergo AI, combining machine learning, symbolic reasoning, and constraint-based logic to enforce consistent, verifiable, and explainable decision-making.
Budget GPT: Optimizing LLM Inference
Built a 'fast + slow' agent framework to reduce overthinking in Large Reasoning Models (LRMs). A Manager LLM predicts an optimal token budget, while a Reflector performs verbal reinforcement learning via natural language error analysis.
Scalable Banking System with Modified Paxos & 2PC
Implemented a fault-tolerant distributed transaction processing system integrating Modified Paxos, sharding, replication, and Two-Phase Commit to achieve strong consistency over 3 replicated shards for 3K+ clients.
Education Statistics & Teaching Quality Metrics
Engineered a data-driven teaching quality metric evaluating 500+ U.S. universities by integrating datasets from RateMyProfessors, College Scorecard, and IPEDS using advanced standardization techniques.
Get in touch.
Interested in collaboration or research opportunities? I'd love to hear from you.
Current Status
© 2025 Saikiran Reddy Jakka. Built with Next.js, TypeScript & Tailwind CSS