Skip to content
View siva-fy's full-sized avatar
🎯
Focusing
🎯
Focusing
5 followers · 4 following

Achievements

Achievement: YOLO

Achievements

Achievement: YOLO

Highlights

Block or report siva-fy

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. Markdown is supported. 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
siva-fy/README.md

Siva — Data and AI Engineer. Complex data. Clear impact.

Portfolio ↗   ·   LinkedIn   ·   Active repositories

Hi, I'm Siva.

I'm Padala Bala Siva Sai Megi Reddy, a Data & AI Engineer based in India. I'm interested in reliable data pipelines, useful backend systems, and the engineering that makes AI applications measurable.

My personal projects explore a connected set of questions: Can the pipeline survive a replay? Can the data be trusted? Can the AI result be evaluated?

Start here

Explore Event Warehouse for a small, replay-safe data pipeline, or Retrieval Eval Lab for an inspectable retrieval baseline. Both run locally with synthetic fixtures:

git clone https://github.com/siva-fy/event-warehouse.git
cd event-warehouse
python3 demo.py
python3 -m unittest discover -s tests -v

Data engineering

Project What you can inspect
Event Warehouse JSONL → raw ledger → validated payments → SQL revenue mart. Deduplication, quarantine, UTC normalization and currency isolation.
CDC Replay Engine Versioned changes, out-of-order updates, tombstones, collision detection and transactional replay.
Data Contract Guard CSV contracts, schema drift, uniqueness, freshness checks and a command-line quality gate.

AI engineering & data for AI

Project What you can inspect
Retrieval Eval Lab BM25 vs. keyword overlap, paragraph citations, recall@k, MRR and an abstention baseline.
LLM Output Evaluator Offline checks for structured responses, citations, abstention, regression coverage, latency and cost estimates.
Point-in-Time Features Historical SQL features that respect event time and data availability to prevent temporal leakage.

Try any of these locally: clone the repository, then run python3 demo.py. Each includes synthetic fixtures, tests, CI, architecture notes and documented tradeoffs. These are AI-assisted learning and reference projects; their demo results are reproducible fixture outputs, not production or model benchmarks.

Earlier explorations

These older projects show my path through ETL, machine learning, web and mobile development. Their READMEs explain setup and current limitations.

Smaller coursework and interface exercises live in the learning archive.

What I work with here

Data & systems: Python · SQL · SQLite · data modeling · ETL · validation
AI foundations: retrieval · evaluation · NLP · TensorFlow · PyTorch
Applications: JavaScript · React · Node.js · Express · GitHub Actions

I like making failure paths visible, keeping examples reproducible, and learning the fundamentals behind the abstractions.

Have an interesting data problem or an idea worth building? Let's connect.

Pinned Loading

  1. event-warehouse event-warehouse Public

    Replay-safe payment event warehouse with validation, quarantine and currency-separated SQL revenue marts. Python + SQLite.

    Python

  2. cdc-replay-engine cdc-replay-engine Public

    Versioned CDC replay with transactional deduplication, out-of-order updates and deletion tombstones. Python + SQLite.

    Python

  3. data-contract-guard data-contract-guard Public

    CSV data contracts with schema drift, uniqueness and reproducible freshness checks. A dependency-free Python quality gate.

    Python

  4. retrieval-eval-lab retrieval-eval-lab Public

    BM25 retrieval with source citations, recall@k, MRR and an inspectable keyword baseline. Offline AI engineering reference.

    Python

  5. llm-output-evaluator llm-output-evaluator Public

    Offline LLM response regression checks: JSON schema, citations, abstention, coverage, reported latency and estimated token cost.

    Python

  6. point-in-time-features point-in-time-features Public

    Leakage-aware historical features using event time, availability time and SQL lookback windows. Python + SQLite.

    Python