AI Advisory Work (2024 Updates)
AI Product/ Advisory Work - 2024
"Your own experience is rarely the whole truth", AI Advisory/Consulting provided me with more collaboration opportunities and focused efforts. The pay cut was worth it. Hoping some equity pays off :). You play multiple roles to get products to market :).
Product Collaboration #1 (April 2024 - Ongoing) - GenAI + Real Estate product focused on solving customer pain points, smooth onboarding, support and sales. RAG, KG, Text2SQL, RDBMS, Redis Cache, Postgres, MongoDB. Multimodal data / multiple scenarios for sales/marketing. Pilot deployments in progress.
Product Collaboration #2 (Sep 2023 - Aug 2024 - Completed) - AI in Building construction domain - AI to solve data, recommendations, Finding right balance of RAG, LLM, NER etc..Building pitch/search implementation / semantic search etc..
Key Responsibilities and Achievements:
Architected an intelligent data catalog leveraging Large Language Models (LLMs) to enhance industrial data organization and accessibility
Led the development of custom LLM fine-tuned and Named Entity Recognition (NER) models specifically tailored for industrial applications
Orchestrated the creation of a comprehensive end-to-end AI engine, integrating various components into a cohesive solution
Conducted in-depth market analysis to identify key opportunities and challenges in the industrial AI space
Contributed to pitch creation, effectively communicating the technical value proposition to potential stakeholders
Ensured the scalability and adaptability of the solution across various industrial sectors, including supply chain, procurement, and data management
Trained the team from Zero AI Skills to built products from multiple experiments/iterations
The team has basic skills using Azure form recognizer, out of box APIs
Key Learnings:
Solved both How to Solve with AI / How to Make Money with AI :)
Key features for VC pitch / Customer Pitch / AI API Strategy / Roadmap
Gained valuable experience in end-to-end AI solution development for industrial applications
Developed a deep understanding of how to apply Deep Learning / GenAI technologies to traditional industrial challenges
Platform Azure, Tech Stack - OpenAI multiple models, VectorDBs / Azure Search etc..
GenAI · LLM · Data Science · Construction · Data Quality
Product Collaboration #3 (May 2024 Ongoing) - Vision-based models to reduce churn in Fashion. This was a very unique use case focused on visual try on before ordering, Sep 2024 go-live planned. Site built feature enhancements in progress - https://fitumi.ai/
Client Project #1 - News / Media (April / May 2024) - Completed - Playbook creation for a news media customer for next best offer. This includes end-to-end strategy of customer analysis, news analysis, interests mapping, recommendation engine, providing previews / recommending subscriptions based on relevant feeds
Client Project #2 - Legal Site Marketplace (April / May 2024) - Completed - Solution Analysis, Product mapping, Customer journey, Lawyer journey, Data mapping, and Knowledge extraction. Phased adoption of use case implementation. Provided product strategy covering Competitor Evaluation, Product Architecture, and Data prioritization.
Client Project #3 - Txt2SQL custom work in Finance domain (May 2024) - Completed - Analyzed use case, Recommended a custom approach which leverages a blend of custom NLP models, Zero shot, few shot LLM approaches—semantic similarity techniques to find reusable queries. Provide end to end approach / incremental benchmark techniques custom based on database.
ML Curriculum Work #1 (May - July 2024) - Completed - Emeritus - NTU - Singapore - ML Assignments / Scenarios / Questions and answers work for 6 modules
GenAI Curriculum Work #2 (July - August 2024) - Completed - Emeritus - ISB Leadership and AI - GenAI Assignments / Scenarios / Questions and answers work for 3 modules and Capstone Projects
ML Curriculum Work #2 (June - November - Completed) - Emeritus - IIMK - GenAI Supply Chain - GenAI Supply chain offerings, examples, adoption recommendations. Reusing my Microsoft Supply Chain into GenAI lens :)
Product Collaboration - Consulting #4 (August 2024 - October 2024) - Completed - Vision-based models for warehouse inventory counting. Analysis - Data Collection - Modelling. Working directly with the CTO to solve the problem. The project closed due to cost issues in Training. Built and Demoed model in GCP Vertex AI. Spent 30K at my own expense as the Customer felt cost was an issue.
Product Collaboration - Consulting #5 (August 2024 Ongoing) - ESG-related data pipelines / Model building. Working directly with Founders to solve the problem. Built a pipeline with a set catalog of items. Currently working on closing on Training / Paused to focus on training. Pipeline testing in progress.
"Everything I work on or choose to work on is a replica of, or closely associated with, my past projects, product exposure, or partial experiments. Every learning adds up and compounds to help me build the first best version."
