CV

VIKTOR TERESHCHENKO

  • Applied AI Engineer
  • Location: Cebu, PH.
  • Languages: English (C2), Russian (Native)

ABOUT

Applied AI Engineer with 3+ years shipping production Large Language Model (LLM) systems — conversational voice agents, Retrieval-Augmented Generation (RAG) pipelines, Model Context Protocol (MCP) servers, and agent evaluation frameworks. Python-first across the full lifecycle: model integration, prompt engineering, async API serving, and deployment on GCP/AWS with Docker and Kubernetes.

TECHNICAL SKILLS

  • Programming Languages: Python, Dart, SQL
  • LLM & Agents: Generative AI (GenAI), NLP, RAG, MCP, A2A, structured outputs, Google ADK, LangChain, Ollama, llama.cpp, Gemini, OpenAI/GPT, Gemma, Stable Diffusion, FLUX.1
  • Serving & Frameworks: FastAPI, FastMCP, Starlette, Flask, Django, Socket.io, Flutter
  • Frontend: JavaScript, Vue, HTML/CSS, Node.js
  • Vector & Databases: PGVector, Qdrant, LanceDB, PostgreSQL, MongoDB, Redis, MySQL
  • ML & Data: Machine Learning, PyTorch, scikit-learn, LightGBM, Polars, Pandas, PySpark
  • Orchestration & ETL: Apache Airflow, Prefect, Taskiq, n8n, Activepieces, Langflow
  • Cloud & DevOps: GCP, AWS, DigitalOcean, Docker, Kubernetes, GitLab CI/CD, Ansible, HAProxy
  • Observability & Testing: Prometheus, Grafana, OpenSearch, JSONL logging, PyTest, pre-commit
  • APIs & Integrations: REST, OAuth, webhooks, gRPC/Protobuf, Stripe, HubSpot, Auth0, ElevenLabs
  • OS: Red Hat Enterprise Linux, Microsoft Windows Server

EXPERIENCE

Software Engineer / Logic Developer
December 2022 – Present
Voctiv Inc. (San Jose, CA, Remote)

  • Built and deployed 200+ production conversational voice AI agents with Python, Google ADK, and LangChain for customers ranging from 10 to 40K calls/day, owning agent logic end-to-end;
  • Designed an internal MCP server (Core / Lite / Meta tool surfaces) that lets AI assistants operate the CMS platform to accelerate agent development and deployment;
  • Built a call-analysis microservice (600 calls/day, cost-capped) exposing a shared FastAPI + MCP surface that transcribes and analyzes call audio with Gemini, backed by Redis caching and task queues;
  • Authored a suite of agent Skills (bootstrap, conformance testing, prompt optimization, call-log analysis) that standardized how the team ships agents with AI coding assistants;
  • Developed an asynchronous LLM agent evaluation framework using LLM-as-judge to grade multi-turn dialogues against requirements and surface concrete prompt fixes;
  • Implemented a RAG question-answering API with FastAPI, MongoDB, and LangChain;
  • Built a multilingual voicemail-detection service with LightGBM, scikit-learn, and PyTorch, reaching 92.3% accuracy (Spanish) and 82.4% (Indonesian);
  • Built a load-balanced TTS inference API with Socket.io, Starlette, and HAProxy;
  • Built ETL pipelines on Apache Airflow and multiple data-processing and automation pipelines;
  • Implemented a Flask/PostgreSQL REST API integrating conversational bot logic with customer CRMs;

PROJECTS

Async Python runtime for a fleet of small, focused AI agents (one container each); multi-LLM backends (Gemini, Vertex AI, OpenAI-compatible), a local MCP tool server, approval-gated actions, and Telegram/Slack/Matrix channels.
Python, Docker, MCP, multi-LLM

MCP server for semantic agent memory — vector search over embeddings, a labeled knowledge graph, and multi-tenant namespaces — exposed as both MCP tools and a REST API.
FastAPI, FastMCP, LanceDB, Gemini

AI document-repair service that iteratively diagnoses and fixes print-layout issues across PDF/DOCX/XLSX/PPTX, with confidence-scored verification, async workers, and MCP tool exposure.
FastAPI, FastMCP, Taskiq, Gemini/Claude

AI-powered text adventures with on-the-fly narrative generation and FLUX.1 image generation.
FastAPI, Flutter, Google Gemini 2.5, FLUX.1, Auth0, Supabase

Environment Monitoring System
Real-time temperature/humidity monitoring. Prometheus, Python, Micropython, and MQTT

TRAINING AND CERTIFICATES

Red Hat Linux

  • RH 133 Red Hat Enterprise Linux System Administration
  • RH 253 Red Hat Enterprise Linux Network Services and Security Administration
  • RHCSA 100148071 Red Hat Certified System Administrator

Apache Airflow, Apache Kafka, Shell

  • KDBSVTNU5OO ETL and Data Pipelines with Shell, Airflow and Kafka
  • SH035Z2M1UVC Data Warehouse Fundamentals

EDUCATION

Bachelor of Science in Automation Systems
Irkutsk National Research Technical University
August 2006 – May 2011