Forward Deployed Engineer
Building AI inside customer workflows, sitting right next to the people who use it. If it doesn't work on day one, it doesn't count.
You're standing on top of the world. Thin air, big dreams.
You slipped off the summit and landed here. Classic. Hi, I'm Mahesh. I build AI that works where it matters. Let me walk you back up, one camp at a time.

Hyderabad. A curious kid with too many questions and one big feeling: that machines could do so much more than they were told.
I didn't start with AI. I started with wires, circuits and "why does this work?" That question never really left.
At Government Polytechnic, Masab Tank, I studied electronics and communication: circuits, signals, microcontrollers. NCC gave me discipline, with B and C certificates on the way.
Then I met machine learning, and everything changed. I started building things nobody asked for: a gas-leak alarm for homes, a bus guide for my city, a chatbot that answers from the Bhagavad Gita.
2023: Python, AI/ML and cloud courses, stacked one after another.
May 2025. I joined Valkontek Embedded & IoT as an intern working on computer vision for cars. No slides, no demos. Real cameras, real vehicles, real latency budgets.
I shipped end to end, from data to deployment on NVIDIA Jetson AGX Orin, and two months later I was an AI Engineer. After that came person re-identification across eight live feeds, and agents my whole team used every day.
YOLOv8/v9/v10, SegFormer, DeepLabV3, UNet, ByteTrack, StrongSORT, Re-ID, OpenCV
RAG, LoRA/PEFT fine-tuning, Unsloth, LangChain, CrewAI, Hugging Face, Gemini
TensorRT, ONNX, Jetson AGX Orin, Docker, FastAPI, AWS (EC2, Bedrock, SageMaker)
PostgreSQL, pgvector, FAISS, ChromaDB, Pinecone, Pandas, Scikit-learn
Building AI inside customer workflows, sitting right next to the people who use it. If it doesn't work on day one, it doesn't count.
ECET lateral entry, studying by correspondence alongside full-time work.