Aug 2024 – Present · 1 yr 8 mos

• Playbooks is a pioneering open-source Software 3.0 framework for creating AI agents using natural language programs. It treats large language models (LLMs) as processors (CPUs) to reliably execute workflows written in English as structured Markdown files. These playbooks are semantically compiled into a low-level assembly language optimized for LLM execution. The framework supports a hybrid execution model where natural language steps and Python code share a unified call stack.
• As the creator and lead developer of Playbooks AI, I envisioned and built all core components, from the semantic compiler and runtime engine to hybrid interop, multi-agent collaboration via the Model Context Protocol (MCP), and VSCode debugging support to step debug through English instructions like traditional code. PlaybooksLM is a specialized small LLM (3B parameters) that is fine-tuned to ensure fast, accurate, secure, and low-cost execution. This empowers non-coders to build complex AI systems, such as conversational agents, automation workflows, adaptive reasoning bots. Advanced capabilities like runtime program synthesis/search, verifiability constraints, multi-agent emergent behavior hint at potential use for AGI.

Aug 2019 – Nov 2024 · 5 yrs 3 mos · (Acquired)

• Senior AI executive leading R&D and product teams in developing enterprise-grade Gen AI solutions, notably creating AutoRAG (advanced RAG platform), TruthChecker (industry-leading hallucination detection system adopted by NVIDIA), and AutoFlows (hypergraph mining for conversational automation). Pioneered multiple RAG techniques 6-18 months ahead of academic publications, including Chain of Thought, Self-Check, and generative reranking.
• Demonstrated success in translating cutting-edge AI research into commercial products, with large enterprise deployments of AI copilots used by 10,000s of customer support and sales agents.

Jan 2017 – Jul 2025 · 8 yrs 6 mos · Redmond, WA

• (on hold) Working on building a new, biologically plausible neural network framework for Artificial General Intelligence, with the goal of building machines that can think and feel like humans by 2030.

Aug 2013 – Jul 2017 · 3 yrs 11 mos · (Acquired by biproxy)

• Led software development, web infrastructure, and product for OfficeSpace.com, a B2B platform for Commercial Real Estate. Amol's team developed a sophisticated ETL infrastructure capable of synthesizing data from a wide range of semi-structured sources. The web platform is utilized by over 100K business users every month.

May 2011 – Jul 2013 · 2 yrs 2 mos · Greater Seattle Area

• Built Edufied, an education technology platform to help students learn through guided repetition. Ragafied was a product built on Edufied where a student learns to sing as if learning from one of the legendary masters. VidoeReadr was another application built on Edufied that allowed educators to build rich media courseware by bringing together video, text and images on a single timeline.

Nov 2010 – Jul 2011 · 8 mos · (Acquired by uZoom)

• Lead technologist behind KingOfTheWeb - a "YouTube with a winner" online reality game that brings together creator economy, social media and reality TV.

Apr 2008 – Nov 2010 · 2 yrs 7 mos · Acquired by Dhingana

• India's first streaming music service. Vision, product, technology and digital media rights.

Apr 2001 – Apr 2008 · 7 yrs

• Helped build several key products and services, including Internet Explorer and Microsoft Office InfoPath, OneNote, Office365 and Bing, resulting in eight patents.

Jul 1997 – Jul 1998 · 1 yr

• Developed B2B collaboration software.