AI Engineer | AI Systems Architect | AI Tech Lead @ 4Good.AI
Focused on designing and deploying production-grade AI systems, with a passion for the intersection of AI, Systems, and Robotics. Currently exploring Rust & Go for high-performance, scalable AI backends.
"There is no AI strategy without a solid data strategy."
Core Expertise:
- Production AI systems architecture and deployment
- Computer vision and NLP
- Multimodal applications and agentic systems
- High-performance AI infrastructure
Current Interests:
- Physical AI and robotics integration
- Scalable AI backends with low memory footprint (Rust, Go)
- Real-time inference optimization
- AI security and evaluation frameworks
Kaio-sight: Autonomous vehicle 7 camera Video AI Pipeline
Built a complete video AI processing pipeline in 4 days on AMD's MI300X platform using NVIDIA's AV dataset.
Technical writeup →
Multimodal Vision-Language Model
Implemented a VLM based on paligemma and siglip from scratch using PyTorch and hugginface tokenizers.
Details →
RAG-based Research Tool
Built an intelligent research assistant using retrieval-augmented generation based on web scraping.
View →
- Springer (2024): Scaling Evaluation of Non-objective Assessments Using AI-Based Solutions
- IEEE (2024): Performance Analysis of Subsea Pipeline Defect Classification Using Supervised ML
- Languages: Python, C++, Rust, Go, C
- MLOps: MLflow, Comet ML
- ML/DL: PyTorch, TensorFlow, scikit-learn, CUDA, ROCm
- Agentic Frameworks: Google ADK, Agno, LangChain, LlamaIndex, LangGraph
- Finetuning & Inference: Unsloth, Hugging Face, vLLM, Ollama
- CV: OpenCV, YOLO, Detectron2, MediaPipe
- Data: Pandas, Polars, PySpark, NumPy
- Infra: Prefect, n8n, Pathway
- X/Twitter
- Medium
- Professional: @Poornachandra102