Looking for a job

Hesam Shams

@hesamshams

Founding ML Engineer at Dwellci AI who took a GCP-native real-estate intelligence platform from zero to paying customers.

United States

Message

At Dwellci AI, I built a GCP-native real-estate intelligence platform from zero to paying customers. I created knowledge-graph RAG over zoning and compliance records and a Rust engine that generates ranked subdivision layouts.

Previously, I built gebra, an open-source toolkit for verifying LangGraph agents, and deployed a computer-vision system that sorted recyclable bottles with 95%+ accuracy. At Sage Intacct, I developed transaction outlier-detection and invoice-field prediction prototypes; earlier, I worked on supply-chain optimization, demand forecasting, and parallel optimization research.

Experience

Work history, roles, and key accomplishments

Current

Founding ML Engineer

Current

Dwellci AI, Inc.

Mar 2025 - Present (1 year 7 months)

Built and operate a GCP-native AI platform for real-estate site discovery and feasibility analysis, growing it from zero to three paying customers. Designed LLM orchestration, GraphRAG, and a Rust geometric-optimization engine.

GCP Cloud Run Cloud SQL PostGIS BigQuery Pub Sub Vertex AI fastAPI Neo4j GraphQL Rust WebAssembly ThreeJS MapboxGL TypeScript React Genetic Algorithms Reinforcement Learning CI CD

Principal ML Engineer

Gebra Tech, Inc.

Apr 2023 - Mar 2025 (1 year 11 months)

Developed open-source developer tooling for AI agents and applied computer-vision client work. Published the gebra package for static verification of LangGraph workflows and deployed a bottle-sorting system on AWS.

Python LangGraph PyPI AWS EC2 S3 Sagemaker MLFlow Computer Vision Open Source

Principal Machine Learning Scientist

Sage Intacct, Inc.

Apr 2021 - Apr 2023 (2 years)

Led new ML initiatives in transaction risk and AP automation for an enterprise accounting SaaS. Prototyped outlier detection, re-architected a legacy rules pipeline into an ML system, and built an OCR invoice-field prediction proof of concept.

Anomaly Detection Neural Networks OCR NLP Classification ML Pipelines Big Data

Senior Data Scientist / ML Engineer

Haulistix, Inc.

Mar 2020 - Apr 2021 (1 year 1 month)

Led a team of six covering data science and analytics at an early-stage supply-chain analytics startup. Built ETL infrastructure and ML pipelines on BigQuery and increased annual revenue by an average of $1M per carrier through simulation-based optimization.

BigQuery ETL ML Pipelines Simulation Optimization Team Leadership

Data Scientist

Zume, Inc.

Nov 2018 - Mar 2020 (1 year 4 months)

Improved demand forecasting accuracy by 25% with a hierarchical framework combining topic modeling, sentiment analysis, ARIMA, exponential smoothing, clustering, and deep learning models. Designed integer-programming and heuristic solvers for greenhouse operations.

Demand Forecasting Topic modeling Sentiment Analysis ARIMA Exponential Smoothing K Means DBSCAN CNN LSTM Gradient Boosting BigQuery Integer Programming Heuristics

Research Assistant

The University of Tennessee

Aug 2014 - Aug 2018 (4 years)

Built an ML-guided parallel optimization solver in C++ on HPC clusters for large-scale NP-hard problems. Developed a Python package for statistical comparison of optimization algorithms and generated infill patterns for additive manufacturing, leading to a US patent.