Samuel P. looks like a good fit?

We can organize an interview with Aldin or any of our 25,000 available candidates within 48 hours. How would you like to proceed?

Schedule Interview Now

Samuel P. - Machine Learning, AWS, nextjs

My name is Samuel P. and I have over 6 years years of experience in the tech industry. I specialize in the following technologies: Machine Learning, TensorFlow, Python, PyTorch, Linux System Administration, etc.. I hold a degree in , Bachelor of Applied Science (BASc). Some of the notable projects I’ve worked on include: Cuelis – Prompt Management Platform for Automation & Agents, AI Customer Support Agent • RAG-Driven Ticket Resolution for Shopify, Vektaris – Decentralized AI-Ready Vector Database, ToolFlow Protocol JSON-First LLM Tool Discovery & Invocation Server, Hybrid Document Intelligence & Simulation Chatbot (MVP), etc.. I am based in Miami Beach, United States. I've successfully completed 27 projects while developing at Softaims.

I employ a methodical and structured approach to solution development, prioritizing deep domain understanding before execution. I excel at systems analysis, creating precise technical specifications, and ensuring that the final solution perfectly maps to the complex business logic it is meant to serve.

My tenure at Softaims has reinforced the importance of careful planning and risk mitigation. I am skilled at breaking down massive, ambiguous problems into manageable, iterative development tasks, ensuring consistent progress and predictable delivery schedules.

I strive for clarity and simplicity in both my technical outputs and my communication. I believe that the most powerful solutions are often the simplest ones, and I am committed to finding those elegant answers for our clients.

Main technologies

  • Machine Learning

    6 years

  • Machine Learning

    3 Years

  • TensorFlow

    5 Years

  • Python

    5 Years

Additional skills

Direct hire

Potentially possible

Ready to get matched with vetted developers fast?

Let’s get started today!

Hire undefined

Experience Highlights

Cuelis – Prompt Management Platform for Automation & Agents

I built Cuelis using TypeScript and React to simplify prompt management for clients running complex automation and agent workflows. Prompts often become difficult to track and update at scale, so I created a platform that allows me and my clients to easily organize, version, and update prompts on the fly—without rebuilding or redeploying the backend. This solution streamlines operations, reduces downtime, and enables fast iteration, making prompt-driven automation more reliable and efficient.

AI Customer Support Agent • RAG-Driven Ticket Resolution for Shopify

Developed an AI support agent that ingests open tickets from third-party sales platforms, normalizes them, and indexes key details into a vector store. Using a Retrieval-Augmented Generation pipeline atop GPT-4, it retrieves relevant FAQ content and historical responses to craft accurate, conversational replies. Integrated via Shopify webhooks and a lightweight React dashboard, the system logs all interactions in Firebase for auditability and continuously refines its knowledge base through periodic retraining.

Vektaris – Decentralized AI-Ready Vector Database

I built Vektaris as an in-house AI embedding storage system, which proved so powerful that I decided to release it publicly. Built on Internet Computer Protocol (ICP), Vektaris provides scalable, secure, and decentralized vector storage for AI-driven applications. 🔹 AI-Optimized Vector Databases – Store, manage, and query embeddings with high-performance retrieval. 🔹 Decentralized & Secure – Blockchain-based immutable data protection and end-to-end encryption. 🔹 Seamless Integration – Python, Node.js SDKs, and APIs for TensorFlow, PyTorch, OpenAI, and more.

ToolFlow Protocol JSON-First LLM Tool Discovery & Invocation Server

ToolFlow Protocol (TFP) is a lightweight, JSON-first spec and reference server—built in Node.js/TypeScript—for discovering and invoking “tools” (APIs) at runtime. • Endpoints: – GET /tfp/tools returns a manifest – POST /tfp/invoke calls a tool by ID • Type Safety: Zod runtime validation + TypeScript types • Extensible: Register new handlers with registerTool(descriptor, handler) • CLI-Friendly: ~150 lines—test instantly via curl Rapidly prototype context-aware toolchains for RAG workflows, custom LLM agents, or human integrations. See my GitHub for examples and setup instructions.

Hybrid Document Intelligence & Simulation Chatbot (MVP)

Built an end-to-end MVP using React & Firebase, OpenAI, and Pinecone to turn uploaded documents into live business simulations. Users chat or drag-and-drop PDFs, DOCX, or XLSX; Firebase handles state and storage. A Python preprocessing pipeline (OCR + format normalization) feeds Pinecone for vector search. OpenAI LLMs extract pricing rules, conditional logic, and parameters, then power a simulation API via FastAPI. Results and human-readable explanations return in a dynamic React interface, ready for rapid iteration and scaling.

Education

  • ITT Technical Institute

    in Information Systems Cyber Security

    2012-01-01-2015-01-01

  • University of London

    Bachelor of Applied Science (BASc) in Artificial Intelligence

    2021-01-01-2026-01-01

Languages

  • English

Personal Accounts