Steve Fischer

AI agents + full-stack

I build agentic systems that turn messy workflows into reviewable outputs.

Full-stack engineer focused on agentic GenAI products, serverless orchestration, and product surfaces that make AI outputs inspectable and useful.

View projectsContact

Current focus

  • Tool-calling agents with explicit context, action, and review boundaries.
  • Serverless orchestration for long-running generation and validation workflows.
  • Product surfaces that make agent outputs inspectable, editable, and useful.

Proof

Agentic project proof

Featured work emphasizes orchestration, tool use, structured outputs, and production-minded guardrails.

Multi-agent reviewPublic source

AI-Enabled Die Hard Damage Assessor

A multimodal, agentic batch pipeline that turns a movie script into a structured “damage assessment” report with evidence-backed findings.

Proof signal

Extract -> validate -> summarize agent workflow

GenAIMultimodalMulti-agenticBatch
  • What I built: A proof-of-concept system for AI-enabled document review: ingest a screenplay and related reference material, chunk it into reviewable units, run multi-agent analysis over text and images, and produce a final report with traceable rationale.
View case studyRepo
Data-to-narrative workflowPublic source

Beat the Brewer Ratings

A GenAI-driven ratings experience that generates spoken recaps and highlights—wrapped in a “Liquid Glass” UI motif.

Proof signal

Ratings -> recap -> voice workflow pattern

GenAITTSUI/DesignProductivity
  • What I built: A lightweight application that turns competition or event ratings into shareable narratives, including TTS output for quick audio recaps and a modern glassy UI style for presentation.
View case studyRepo
Transcript extractionPublic source

Meeting Transcript Analyzer

A productivity pipeline that converts meeting transcripts into structured summaries with speaker attribution and actionable takeaways.

Proof signal

Decisions + actions + open questions artifact produced

GenAIText ProcessingSTTSpeaker Identification
  • What I built: A GenAI text-processing system that ingests meeting transcripts, identifies speakers, chunks the discussion into coherent segments, and produces summaries, decisions, and action items suitable for sharing.
View case studyRepo

Signals

What I ship

Stack

Next.js, TypeScript, Tailwind

Agents

Tool use, orchestration, evaluation

Infra

Serverless AWS, Step Functions

Next

Let's build something useful

I partner with teams that need agentic prototypes hardened into reliable product workflows.

Have a GenAI product to ship?

I can help with agent architecture, implementation, evaluation loops, and delivery for full-stack experiences.

Start a conversationExplore work
Based in the US. Open to remote collaboration.More about me