Hi, I'm Jaynil Jaiswal
Software Developer | ML Engineer
Specializing in AI Infrastructure, Orchestration, and Data Engineering.
About Me
I'm a Software Developer and Machine Learning Engineer with a focus on building robust AI control planes, orchestration platforms, and scalable data infrastructure. My expertise spans from low-level systems programming in C++ and Go to architecting distributed microservices and RAG pipelines.
Education
- 🎓M.S. in Computer Science
UC San Diego, 2023 - 2025 - 🎓B.Tech in Computer Science
Indian Institute of Technology Roorkee, 2021
Certifications
- 📜IBM Data Engineering Professional
Issued Apr 2025
Skills
Featured Projects
LLM-Powered Search Engine
A custom search engine powered by large language models that allows natural language queries across my portfolio projects and experience.
Durable Multi-Agent Orchestration
Architected a fault-tolerant agent orchestration platform using Temporal.io, enabling durable workflows, stateful orchestration, and 'Skeletonizer' memory management.
Matrix Multiplication on GeneSys
Programmed GeneSys hardware with a 4x4 systolic array of PEs for efficient matrix multiplication, leveraging iterator tables and instruction scheduling.
Sales Data Warehouse & BI
Designed a scalable OLAP Data Warehouse in PostgreSQL. Built ETL pipelines handling 10M+ daily records for a Looker Studio dashboard.
Enterprise AI Gateway
Built a high-performance AI Gateway using KServe for model serving and Redis for semantic caching. Reduced query latency by 60%.
Optimized Matrix Multiplication
Engineered a highly optimized SGEMM kernel in CUDA, achieving 3386 GFLOPS. Utilized 2D shared memory tiling and coalesced memory reads.
Distributed Cache with gRPC
Built a high-availability distributed in-memory cache in Go. Implemented custom Gossip protocol and gRPC bidirectional streaming.
Instruction-Following LLM Fine-Tuning
Fine-tuned FLAN-T5-XL and Mistral-7B on Bitext dataset using LoRA, achieving 28% ROUGE-L improvement with 60% memory reduction.
Streaming ETL Pipeline
Automated a scalable real-time ETL pipeline with Apache Kafka, MySQL, and Airflow for toll traffic analysis, handling 1M+ records.
Professional Experience
Graduate Student Researcher
Apr 2024 - PresentWiFire Lab, SDSC, UCSD
San Diego, CA
- Architected a High-Throughput Data Pipeline for a 234GB spatiotemporal dataset, optimizing ingestion for Transformer (ViViT) models.
- Optimized Computational Efficiency by implementing O(N log N) Fourier-domain mixing algorithms.
- Engineered a 'Differentiable Simulation' Framework (Soft-ROS), integrating physical constraints directly into the Gradient Flow.
- Standardized Scientific Workflows using MLflow, establishing a robust Model Registry.
Software Engineer IC2
July 2021 - Sep 2023Oracle
Bangalore, India
- Scaled and optimized the Distributed Control Plane for a global Dynamic Routing System across 18+ regions.
- Engineered a 'Bandwidth Broker' POC using Aerospike and Redis to process 1M+ metric updates/sec.
- Developed the 'Fleet Health Visualizer,' a Go-based monitoring tool utilizing real-time telemetry.
- Engineered a 'Developer Control Plane' through standardized SDKs and Terraform providers.
MLOps Engineer
May 2020 - Feb 2021Proton AutoML (Acquired by Cliently)
Remote, US
- Led the architectural evolution of core AutoML services into a resilient Microservices Architecture.
- Architected a cost-optimized, self-managed Kubernetes platform for high-density ML workloads.
- Developed Custom Kubernetes Operators (CRDs) in Go to automate the lifecycle of training jobs.
- Hardened the Inference Data Plane by migrating to a high-concurrency architecture using Nginx and AWS ALBs.
Latest Insights
MLOps Platforms: The 2025 CTO’s Guide
Introduction — The True Cost of Technical Debt in Machine Learning...
Read MoreBuilding an Efficient RAG System
The ability to efficiently search and retrieve relevant information from massive datasets is a cornerstone of modern AI...
Read MoreRAG Isn’t Dead, It’s Evolving
If you’re building with AI, you know that Large Language Models (LLMs) are incredibly powerful. But you also know their limitations...
Read MoreAI-Powered Portfolio Search
Explore my AI-powered portfolio search engine — ask questions and discover my projects interactively using my custom Serverless RAG Gateway.
Hi! I'm the AI Gateway for Jaynil's portfolio. Ask me anything about his experience engineering AI Infrastructure and Distributed Systems.
Get In Touch
Let's collaborate
I'm always interested in hearing about new projects, research opportunities, or just chatting about AI and software engineering.