Headline

AI Software Engineer | GenAI | Forecasting | Agentic Pipelines & Computer Vision | Machine Learning | Deep learning | Building Products

About

I didn't start in AI because it was a trend; I was drawn to it by a fundamental need to understand how complex systems work from the inside out. For me, fixing a problem means getting to the root cause rather than applying a temporary patch. Holding myself to that standard has defined my approach from day one, and it remains the core driver of my work. Over the past two years, that mindset has pushed me to tackle ambitious, unstructured challenges from developing long-term forecasting models and embedding computer vision into physical hardware to engineering cold-start agricultural intelligence pipelines. Growth hasn't always been linear. I’ve learned that technical accuracy is only half the equation; clear communication, alignment with stakeholder intent, and building user trust are what ultimately determine a system's real-world impact. Ownership of those lessons has made me a far more pragmatic engineer. Looking ahead, my focus is on deepening my expertise in agentic workflows, advanced retrieval, and multi-model orchestration. The most transformative AI solutions won't just perform well on benchmarks; they will seamlessly bridge messy, real-world data with human decision-making. That is the domain I am committed to mastering over the long term. I approach this field with humility and speed. I value intellectual honesty, adapt rapidly, and thrive in environments tackling unsolved problems without a predefined playbook. If your team is building in that space, I’d welcome the chance to connect.

Location

Bengaluru

AI stack

antigravity

claude

cursor

hermes agent

langchain

mcp

Work experience

AI Software Engineer at eAge Technologies India Pvt Ltd, Jul 2025 - Present · 1 yr 3 mos, Bengaluru, The mission: eAge builds industrial data-logging hardware. My role is turning that raw logged data into full AI platforms forecasting, computer vision, agentic decision systems with minimal senior technical oversight. Built and benchmarked LightGBM and LSTM forecasting models for solar power output using 12 years of location-based historical data, reaching evaluation scores of 89% and 80% respectively; designed retraining triggers and performance ratio-based drift monitoring to keep the system reliable over time. Architected an agentic decision pipeline integrating LightGBM/LSTM forecasts with rule-based risk logic and LLM function calling, generating real-time operational directives and automated risk alerts for solar assets. Built a real-time computer vision pipeline (YOLOv8, InsightFace, FAISS) for face detection, distance estimation, and object detection, integrated with STM32 hardware to deliver an end-to-end hardware-software system for defense application. Designed an IoT-to-cloud Agri-Soil Intelligence engine under zero-historical-data constraints, engineering cold-start threshold logic and location-based climate API pipelines across 2 pilot locations. Served as the primary technical interface to stakeholder and engineering leads, translating complex operational requirements into technical specifications. Conducted applied R&D across predictive analytics, computer vision, and forecasting, evaluating model performance through structured experimentation to guide architecture and solution decisions.

Tech Intern at Medha Tech Solutions, Mar 2025 - Jun 2025 · 4 mos, Bengaluru, Karnataka, India

Artificial Intelligence and Data Science Intern at Code Nimbus Solutions, Feb 2025 - Jun 2025 · 5 mos, Bengaluru, Karnataka, India

Frontend Intern at Code Nimbus Solutions, Sep 2024 - Feb 2025 · 6 mos, Bengaluru, Built AI-powered facial recognition pipelines using deep learning and embedding extraction techniques. Developed user interfaces for AI applications, enabling visualization of model outputs and operational workflows.

Education

Bachelor of Engineering - BE, Artificial Intelligence and Data Science, SDM INSTITUTE OF TECHNOLOGY (SDMIT), UJIRE

Public links

github: https://github.com/sangamgowda