Skip to content
View jay739's full-sized avatar

Block or report jay739

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
jay739/README.md
Jayakrishna Konda

Building production RAG pipelines, LLM systems, and self-hosted AI infrastructure, from model to deployment.

Quick nav: About · Batcave · Projects · Releases · Highlights · Blog


Portfolio LinkedIn Email Blog Profile views


Batcave live status


About

I'm an ML/AI Engineer & Data Scientist, and this GitHub is where my projects live: RAG pipelines, LLM systems, and self-hosted AI infrastructure. I build them end to end, from model training and LLM fine-tuning to deployment and real-time monitoring.

The biggest one is Batcave, my solo-built, 56-container private AI server running local LLMs, RAG, and ML inference at 99.9% uptime (live status).

education: M.S. Data Science, UMBC (GPA 3.91 / 4.0)
focus: LLMs · RAG · MLOps · Self-hosted infrastructure
homelab: 56 containers · 36+ services · 99.9% uptime
based_in: Maryland, USA

Open to

  • Full-time ML/AI or MLOps engineering roles.
  • Contract and advisory work for production RAG/LLM systems.
  • Collaborations around self-hosted AI, homelab observability, and applied GenAI.

Batcave: My Private AI Cloud

A solo-managed homelab on a Beelink mini-PC. Self-hosted from scratch. Live at jay739.dev.

Metric Value Metric Value
Containers 56 Self-hosted Services 36+
Uptime 99.9% Local LLMs 5+ models
Hybrid-cloud Latency −60% Multi-DB Layer 177 GB

Stack: Docker · Ollama · LangChain · Authentik SSO · Tailscale · Netdata · Nginx · PostgreSQL · MariaDB · Redis · Meilisearch

Proof links: Live Batcave status · Batcave architecture series · OCI migration + latency notes

→ Read the full 5-part Batcave blog series


Tech Stack

Core: Python · PyTorch · LangChain · Ollama · Docker · AWS · PostgreSQL · Redis · Next.js · TypeScript

Apps: Swift · Kotlin · Dart · Flutter · C# · Go

Full badge stack

ML / AI

Python PyTorch TensorFlow scikit-learn LangChain Ollama HuggingFace PySpark

MLOps · Infra · Cloud

Docker Kubernetes AWS Oracle Cloud GitHub Actions MLflow Tailscale Nginx

Data

PostgreSQL MariaDB Redis Pandas NumPy Plotly

Frontend & Edge

Next.js TypeScript Tailwind ESP32 C++

Native Apps & Services

Swift Kotlin Dart Flutter C# Go


Featured Projects

Solo-built private AI infra on a Beelink mini-PC. 56 containers, 36+ services, local LLMs via Ollama, RAG pipelines, SSO, full observability.

Docker · Ollama · LangChain · Authentik · Tailscale

PDF → LLM → TTS pipeline that turns books into character-voiced podcasts. PyMuPDF extracts the text, FAISS retrieves the passages, Llama 3.1 8B writes the script through Ollama, and Bark voices each speaker.

RAG · FAISS · Ollama · Bark TTS · Electron

PySpark ML pipeline on IBM's 32M-row HI-Medium AML dataset, with temporal feature engineering at a 0.23% laundering rate. The Random Forest baseline reaches 99.9% accuracy but only ROC AUC 0.50, so the repo reports it as chance-level and lists the fixes.

PySpark · MLlib · Random Forest · Financial AI

Benchmarked 8 models on the FLAME aerial dataset in RGB, thermal IR and RGB+IR fusion. For the pretrained backbones, fusion and RGB alone finished within one test image of each other. Best: ResNet-18 on RGB at 99.8% test accuracy, on a 1% subset (about 107 test images).

CNN · Computer Vision · Aerial Imagery · Thermal IR

  Histopathologic Cancer Detection

Compared custom CNN vs ResNet18 / VGG16 / EfficientNet on 220k+ pathology patches. Transfer learning best at 0.94 AUC.

PyTorch · Transfer Learning · Medical AI

Streamlit dashboard for crisis-period portfolio risk: 95% VaR, Monte Carlo (GBM) simulation and FinBERT news sentiment, with backtesting windows for the 2008 crash and the 2020 COVID-19 crash.

FinBERT · Monte Carlo · Streamlit · yFinance · Plotly

→ See more projects on jay739.dev


Latest Releases

  • openhaze v0.3.0 · Native, dependency-free window-dimming utility for macOS and Windows, inspired by HazeOver Nov 9, 2026
  • claude-pop-quiz v0.5.3 · Mandatory, automatic pop-quiz hook for Claude Code. It quizzes you on your own sessions every 40 to 45 actions to counter agentic context loss. Oct 27, 2026
  • sitrep v0.1.0 · Self-hosted status page over Uptime Kuma: incidents, RSS, Telegram notifications Sep 30, 2026
  • cadence v0.1.0 · Per-song music request service: search, request, and auto-fulfil single tracks into a self-hosted library Sep 25, 2026
  • omnifin v1.4.1 · Jellyfin user management, a redesigned hard fork of hrfee/jfa-go Sep 25, 2026
  • sitediff v0.1.0 · Compare the visible content of two renderings of the same page, across environments. Aug 19, 2026

GitHub Activity

Metrics



GitHub streak
More: coding habits & topics
Coding habits

Topics

Achievements

Featured repositories metrics

Notable contributions

Contribution Snake

Contribution snake

Rendered by my fork of Platane/snk, which I extended so the snake grows as it eats.


Latest from the Blog


• Benchmarking Local LLMs Across an RTX 3060 Ti and an M4 Mac Mini (With a Kernel Panic Along the Way) Jul 16, 2026
• The Freeze Wasn't Memory: Tracing a Homelab Server's Root Cause Through Three Wrong Turns Jul 16, 2026
• I Made My AI Ambush Me With Pop Quizzes May 29, 2026
• The Quiet Math of Lifetime Software Deals May 12, 2026
• What Running a Homelab Actually Teaches You That Work Doesn't Apr 24, 2026

→ More at jay739.dev/blog · auto-refreshed every 6 hours from jay739.dev/rss.xml


Let's build something.

"From model to deployment, and the homelab in between."

footer

Pinned Loading

  1. openhaze openhaze Public

    Native, dependency-free window-dimming utility for macOS and Windows, inspired by HazeOver

    Swift 19 2

  2. claude-memory-checkpoint claude-memory-checkpoint Public

    A Claude Code hook that summarizes a long chat into project memory every N tool calls, so parallel/future chats load the summary instead of re-deriving.

    Python

  3. claude-pop-quiz claude-pop-quiz Public

    Mandatory, automatic pop-quiz hook for Claude Code. It quizzes you on your own sessions every 40 to 45 actions to counter agentic context loss.

    Python

  4. omnifin omnifin Public

    Forked from hrfee/jfa-go

    Jellyfin user management, a redesigned hard fork of hrfee/jfa-go

    Go 1 1

  5. scribe scribe Public

    Self-hosted, GPU-accelerated transcription with speaker labels. An open-source Otter.ai alternative: faster-whisper large-v3 + pyannote diarization, web UI, API and CLI.

    Python 1

  6. wakeharder wakeharder Public

    Open-source Android alarm app that only disarms after you complete a task: solve a math problem or scan a pre-registered barcode/QR. Kotlin + Jetpack Compose.

    Kotlin 1