Skip to main content

Tech stack

What we work with, and what we do with it.

A stack list on its own proves nothing. These are the tools we reach for, grouped by the problem they solve, and every one of them appears somewhere in the products we've shipped.
Technologies
39
Areas
6
Products shipped
20
  • 01

    AI & machine learning

    Model training, fine-tuning, and inference — from transformers to diffusion and on-device CNNs.

    • PyTorch
    • Hugging Face Transformers
    • PEFT / LoRA
    • LangGraph
    • Ollama
    • ONNX Runtime
    • scikit-learn
    • OpenCV
  • 02

    Retrieval & data

    Vector search, hybrid retrieval, and the storage layers underneath them.

    • pgvector
    • Qdrant
    • FAISS
    • PostgreSQL
    • Redis
    • Kafka
  • 03

    Backend & APIs

    Typed, documented services built to be called by something other than a demo script.

    • FastAPI
    • Python
    • Node.js
    • TypeScript
    • Go
    • gRPC
    • WebSockets
  • 04

    Frontend

    Accessible, fast interfaces — server-rendered by default, interactive where it matters.

    • Next.js
    • React
    • TypeScript
    • Tailwind CSS
    • Framer Motion
    • Flutter
  • 05

    Infrastructure & MLOps

    Containerized services, reproducible deploys, and the observability to know they work.

    • Docker
    • Kubernetes
    • GitHub Actions
    • Terraform
    • Prometheus
    • Grafana
    • AWS
    • Vercel
  • 06

    Quality & evaluation

    Tests for the software, evals for the models. Both run in CI.

    • pytest
    • Playwright
    • ESLint
    • RAGAS-style evals
    • Load testing

How the work runs

Every engagement follows the same four steps.

  1. 01

    Discovery

    We start with your problem and constraints — what success looks like, and what it will take to get there.

  2. 02

    Plan & Estimate

    A clear scope, milestones, and a realistic timeline, so you know what you're getting and when.

  3. 03

    Build & Deploy

    Iterative delivery with working software at each step — containerized and shipped, not left in a notebook.

  4. 04

    Support & Iterate

    Once it's live, we monitor, maintain, and improve it under real-world use.

Let's talk

Have a problem worth solving?

Tell us what you're building. We'll help you scope it, build it, and ship it.