I build high-performance backend and AI systems — async runtimes, multi-modal inference engines, real-time forensics pipelines. Currently at PaladinAI where I led development of deepfake detection systems, phonetic transcription/translation services, and GPU-accelerated inference infrastructure using Kafka and Flink.
I write production Rust every day. I care about throughput, correctness, and systems that don't fall over.
Interests: AI for Forensics · Blockchain Architecture · Distributed Systems · High-Performance Computing
PaladinAI — Software Engineer Feb 2025 – Present
- Led deepfake detection & AI forensics system — scalable architecture, database design, secure sensitive-data handling
- Built backend services in Rust, Node.js, FastAPI, and PostgreSQL
- Led Phonetic Systems for audio transcription and translation at scale
- Optimized AI inference with GPU acceleration, Kafka + Flink task queue architecture
PaladinAI — Backend Developer Jul 2024 – Nov 2024
- Real-time KYC system: frame extraction, identity verification, blacklist checks, face-swap detection
- Microservice architecture with high throughput and low-latency AI workloads
- Rust + Tokio async services, SQLx type-safe queries, FFmpeg for real-time frame extraction
Web3sec.news — Open Source Contributor Jan 2024 – Dec 2024
- In-depth technical blogs on Ethereum, Solana — consensus models, architecture, performance
- Built a decentralized Ethereum crowdfunding platform: Solidity smart contracts, Ethers.js, React.js
- Wallet integrations with secure on-chain fund management
Scalable async microservice for multi-modal AI inference — Video, Audio, Image — with hardware-accelerated processing.
- High-concurrency API with Poem + Tokio, handling up to 500 MB multi-part uploads and long-running deep learning tasks
- Integrated Torch (tch-rs) and OpenCV for video frame and audio stream processing — deepfake detection and media analysis
- Dynamic device selection: CUDA (NVIDIA) → MPS (Apple Silicon) → CPU fallback, automatic per environment
- Atomic background worker pattern with SQLite for task lifecycle management (
Pending → Processing → Complete), heartbeat monitoring, zombie-task recovery - Async progress reporting via MPSC channels with frame-accurate status via RESTful endpoints
- Strictly typed OpenAPI/Swagger spec for model discovery and frontend integration
Rust · Torch (tch-rs) · Safetensors · OpenCV · Poem · Poem-OpenAPI · Tokio · SQLx · SQLite · CUDA/MPS/CPU
Production fraud detection system with real-time identity verification at scale.
- Frame extraction pipeline via FFmpeg, identity verification, blacklist checks, face-swap detection
- Tokio async runtime + SQLx type-safe queries for reliable, low-latency operations
- Efficient schema design and indexing for high-throughput AI inference workloads
Rust · Tokio · SQLx · PostgreSQL · FFmpeg
Decentralized crowdfunding platform on Ethereum with secure on-chain fund flows.
- Smart contracts in Solidity, frontend via Ethers.js + React.js
- Wallet integration, transparent fund management, safe transaction flows
Solidity · Ethers.js · React.js · Ethereum
| Languages | Rust · JavaScript · Node.js · Python · Solidity |
| Async / Runtime | Tokio · async-std · Poem |
| AI / ML | Torch (tch-rs) · Safetensors · OpenCV · CUDA · MPS |
| Data | SQLx · PostgreSQL · MongoDB · SQLite |
| Streaming | Kafka · Apache Flink |
| Web / API | FastAPI · Express.js · OpenAPI/Swagger |
| Web3 | Solidity · Ethers.js · Ethereum · Solana |
| DevOps | Docker · AWS · Git · Linux |
B.Tech — ABESIT, Ghaziabad Graduated Aug 2024
unsafe-free block at a time.
fn current_focus() -> Vec<&'static str> {
vec![
"Distributed Systems",
"AI Infrastructure",
"Rust Backend Engineering",
"Kafka & Flink",
"High Performance APIs",
]
}