Ambuj Kumar Tripathi

Evaluate and annotate multimodal AI model outputs for RLHF and generative AI video training pipelines, achieving a 97% lifetime quality score and a single-day peak of 1,112 annotations.

Fine-tuned six domain-adapted Indian Legal Llama models using QLoRA and published GGUF quantizations on Hugging Face, and built LLM-as-a-Judge evaluation frameworks for production agentic RAG pipelines.

Engineered constraint-based prompt architectures for commercial image generation models, executed iterative prompt engineering for brand assets, and conducted adversarial red-teaming to detect failures in generative AI outputs.

Developed AI-assisted automation prototypes using OpenAI APIs and Gemini Vision for network documentation workflows, built an NLP query interface, and architected Python data pipelines reducing manual efforts by 60%.

Optimized ISP infrastructure with Orion monitoring, reducing downtime by 10%.

Executed fiber optic deployment projects for the Ministry of Defense across three states, ensuring complete security compliance.

Managed revenue systems for Delhi Jal Board, ensuring 99% uptime through proactive monitoring.

Post Graduate Diploma in Power Transmission and Distribution from National Power Training Institute (NPTI) from 2015 to 2016.

Bachelor of Technology (B.Tech.) in Electrical and Electronics Engineering from Uttar Pradesh Technical University (UPTU) from 2009 to 2013.