Ilia Likhanin

AI Engineer · C# / .NET, Python, language models

22 years in IT. For the last 8, software developer at a medical center: more than 10 programs and services that run every day. I build systems end to end — from the problem statement to production and support.

  • Tashkent
  • Remote or relocation to the US
  • 22 years in IT
Резюме PDF · RUResume PDF · ENTwo pages. The file lists e-mail and the site address.

Employers aren't named on this site: the clinic and the retail chain are described without names on purpose. Company names, references and the contacts of people I worked with: in conversation, on request.

In short

In IT since 2004. Thirteen years at a retail chain: I started as an assistant system administrator and grew into the head of IT while the chain grew from 5 to 67 stores.

Since 2018, software developer at a medical center. I built and rolled out more than 10 programs and services in C# / .NET, Python, Firebird and Excel/VBA: call speech analytics on language models, lab results integrated into the medical system with built-in AI agents, tracking of paper medical charts with long-range RFID (UHF), IP telephony with reports, electronic outpatient records, and a building safety system on microcontrollers.

On contract, I delivered 1C 8.3 projects for a fuel company: real-time exchange with discount card processing, after which the gas station chain moved from 1C 7.7 to 1C 8.3.

I run language models on our own hardware, so the data never leaves for the cloud. Since 2026 I develop together with AI agents: Claude Code, Cursor, Codex.

For a US team
  • Time zoneTashkent, UTC+5. Your morning is my evening: 9:00–13:00 in New York is 18:00–22:00 here, and I keep those hours free for calls. With California the overlap is 9:00–11:00 their time.
  • VisaNo US work authorization yet: I will need visa sponsorship. Ready to start remotely on a contract and relocate once the visa is through.
  • EnglishI read and write English every day; interviews in English are fine. Russian is my native language.
In numbers
  • 25,000+calls transcribed and analyzed by speech analytics — the director saw where the clinic was losing patients
  • 4 h → 30 mina day of staff time on lab results after the module was integrated into the medical system
  • 20–25kmedical charts tracked with long-range RFID
  • 30 dayswithout a failure — a language model on old graphics cards
  • 24 roomswith panic buttons: the clinic received its building safety certificate
  • 5 → 67stores in the retail chain where I grew into the head of IT
Experience
2018 — 2026 · Medical center

Software developer

On staff until 2022, then on contract. More than 10 programs and services around the medical database:

  • Call speech analytics: speech recognition (GigaAM), Gemma and Qwen language models on the clinic's own servers, anonymization, each call analyzed on 36 parameters. 4,000–6,000 calls a month, more than 25,000 in six months.
  • A lab results module integrated into the medical system with built-in AI agents: staff time on lab results is 30 minutes a day instead of 4 hours.
  • Tracking of 20,000–25,000 paper charts with long-range RFID (UHF): 5 readers, 12 antennas, IP camera snapshots, an Android app for finding a chart.
  • “Electronic Nurse” — an AI assistant for doctors, a pilot project: records the appointment, drafts the medical documentation, shows past appointments on one screen.
  • Language models on our own hardware: a custom llama.cpp build for NVIDIA Pascal GPUs, a task balancer in Python, 30 days of work without a failure.
  • A building safety system on microcontrollers (AVR, ESP32): panic buttons in 24 rooms, 8-hour emergency lighting, voice alerts. The clinic received its building safety certificate.
  • IP telephony and call reports, electronic outpatient records, tracking of regulated medications, management reports in Excel/VBA.
2023 — 2024 · Fuel company

1C 8.3 projects

On contract. Designed and built real-time exchange between discount card processing and 1C 8.3 — the gas station chain moved from 1C 7.7 to 1C 8.3 step by step. Extended the 1C 8.3 configuration for the company's car service.

2004 — 2018 · Retail chain

From assistant sysadmin to head of IT

Assistant system administrator → system administrator → head of a three-person IT department.

  • IT setup for every new store while the chain grew from 5 to 67: computers, cash registers, scanners, scales, video surveillance.
  • Infrastructure of two offices, about 180 computers: Windows Server, Hyper-V, backups, 1C 7.7–8.3, MS SQL Server, Firebird, PostgreSQL.
  • A 1C data exchange program over several independent channels and monitoring of 1C services and retail platforms.
  • Networks and seamless Wi-Fi in the offices and stores.
Skills and stack

Development

  • C# / .NET 10: ASP.NET Core, WPF, WinForms
  • Python
  • Java (Android)
  • SQL
  • 1C:Enterprise 8.3
  • Excel/VBA
  • C/C++ for microcontrollers (AVR, ESP32)
  • Next.js, TypeScript

AI and language models

  • Gemma and Qwen on llama.cpp (custom builds)
  • Claude via OpenRouter
  • GigaAM speech recognition
  • data anonymization
  • structured JSON output
  • Claude Code, Cursor, Codex

Databases

  • Firebird
  • MS SQL Server
  • PostgreSQL
  • SQLite

Infrastructure and hardware

  • Windows Server, Hyper-V
  • IP telephony (SIP)
  • networks and Wi-Fi
  • UHF RFID
  • GSM gateways
Education
  • 2023 — 2027Bachelor's in Information Systems and TechnologiesSynergy University, part-time · graduating in March 2027
  • 2023Professional retraining: Software Development Technology, 822 hoursMUTs DPO TsPP training center, Moscow
  • 2020Professional retraining: System Administration and Information Technology, 520 hoursMUTs DPO TsPP training center, Moscow
  • 20161C:Professional certificateand courses in configuring and administering 1C:Enterprise 8, 2010–2016
  • 2005Vocational school, electrician