Klein – Python, OpenAI, LLM

It's a specialized speech-to-text SaaS offering fast, accurate transcriptions via custom ASR models for students and researchers.

• Created a full-stack SaaS audio transcription platform using Wasp (Node.js/Prisma) and a dynamic front end using React;
• Integrated third-party services—Google Analytics, Auth0, and Stripe—for user insights, secure authentication, and payment management;
• Created a real-time highlighting system syncing audio with transcript text for smoother editing and interactivity;
• Oversaw the product design process, establishing information architecture, visual design (UI/UX), and user journeys to produce a user-friendly interface;
• Implemented Whisper-small ASR on AWS SageMaker for a fast, cost-effective transcription option in the free tier;
• Carried out data engineering procedures, such as cleaning, transforming, and preparing for AI model training and application use, on proprietary datasets;
• Created an automated post-processing pipeline to improve the grammar, punctuation, and intelligibility of ASR transcripts by utilizing OpenAI's GPT-4o API.

It's a tool that delivers real-time, context-specific behavioral coaching in Outlook, Teams, Gmail, and Slack to scale company culture.

• Optimized AI pipelines for HR coaching, cutting latency by 62% via prompt tuning, step reduction, and smart caching;
• Reduced infrastructure costs by 30% and boosted accuracy by 20% via efficient token use, strategic caching, and varied AI tools for HR coaching;
• Designed evaluation pipelines and synthetic data workflows to speed up QA and delivery for HR coaching used by 10k+ users per client.

It's a tool that provides professional-grade AI technology for equity research to the investment banking, hedge fund, asset management, and wealth management industries.

• Designed and implemented an advanced AI pipeline for financial document processing, integrating multi-agent systems, RAG architectures, and prompt engineering techniques, directly contributing to a $5M pre-seed valuation;
• Deployed full-stack AI infrastructure on AWS, scaling from R&D to production-ready MVP;
• Served as Interim CTO for 10 months, scaling the engineering team from 2 to 7 members;
• Developed comprehensive evaluation frameworks with custom benchmarks to measure the performance of agents, prompts, and RAG systems;
• Established quality assurance processes and documentation standards for maintaining technical advantages through research-backed approaches;
• Successfully presented to VCs and prospects, helping secure initial funding and establish early customer relationships;
• Transformed the product into a fully operational GenAI-based financial analysis tool with faster insights, reliable data aggregation, and an intuitive UX.

It's a personalized AI writing assistant that helps writers organize work, access their library, and receive tailored creative support with AI-powered features.

• Developed a proof-of-concept for a chat-with-your-data copilot to assist writers with personalized content retrieval and creative support;
• Implemented an agent system architecture for specialized writing assistance tasks, including brainstorming, editing, and research;
• Utilized advanced RAG techniques to enhance contextual understanding of user documents and writing style;
• Integrated vector databases for efficient semantic search across personal document libraries and writing collections;
• Leveraged LangChain and LlamaIndex frameworks to build modular, extensible components for document processing and conversation management;
• Designed intelligent prompt engineering workflows to optimize agent responses based on user writing preferences and goals.