Python Developer / AI Engineer

1. PROJECT

After handover, the client receives the entire system and manages it independently.

2. WHAT THE SYSTEM DOES

• Searches open sources: scientific databases (ISTINA MSU, eLibrary, OpenAlex, arXiv), professional communities and portfolios (GitHub, ODS, Kaggle)
• Works with closed platforms (hh.ru, Habr Career, LinkedIn) using the client's accounts, manually and without mass downloads
• Studies candidate profiles and works, evaluating them based on criteria set by the client
• Maintains a candidate database in Excel according to the client's structure: profile, assessment, priority, points to confirm during an interview
• Launches on client request. The client changes criteria and sources independently, without a programmer
• Does not contact candidates: only searches and evaluates. The client contacts candidates themselves

3. WHAT THE CLIENT PROVIDES

• Candidate database spreadsheet template (Excel), which the system fills automatically: this shows the desired result
• Candidate search documentation: where to search, whom to search for, by what criteria to evaluate, how to work with platforms — for system setup
• The template and documentation are provided after signing a non-disclosure agreement

5. WHAT THE CLIENT RECEIVES

• Source code in its entirety — becomes the client's property
• All access (server, API keys, accounts) registered to the client
• Instructions: launch, changing criteria and sources, troubleshooting common issues
• Independence: the system operates without contractor involvement

To build an AI-based system that independently searches for candidates for the position of a quantitative engineer (hypothesis testing and verification, data cleaning and accounting, Python), studies their work, and maintains a candidate database.

4. TECHNICAL FRAMEWORK

• Preferably based on Claude (Anthropic) models; final choice is discussed with the contractor

• Operation on a cloud server

• Candidate personal data is stored only in the client's system; personal data protection laws are observed

• Experience in building AI agents using language model APIs, preferably Claude

• Experience in collecting data from open sources and working with their APIs; understanding of platform limitations on automation

• Python and cloud server deployment

• Careful handling of personal data

• Portfolio of similar projects; turnkey delivery with documentation

• Willingness to sign a non-disclosure agreement (NDA)

Location: Moscow, Russia Employment: Full-time Salary: 150000–200000 RUB per month

One-time order. The contractor proposes cost and deadlines. Each phase is paid only after client acceptance (or via a secure transaction service).

Phase 0. Preparation — no payment

• Non-disclosure agreement and service contract; exclusive rights to the code transfer to the client
• Architecture: the contractor describes in writing how the system is structured and operates — its components, where and in what format data is stored, how the search process works from launch to filling the table
• System quality criteria: how it's measured that the system searches and evaluates well (e.g., percentage of suitable candidates, completeness and accuracy of table filling, absence of duplicates, compliance with platform limitations)
• Work completion criteria: what exactly needs to be done and verified for the acceptance of each phase
• Architecture and both sets of criteria are approved by the client before work begins and become an appendix to the contract
• Server, API keys, and code repository are immediately registered to the client; the contractor receives access but does not own them

Phase 1. Prototype — approx. 25%

• Search from 1-2 open sources, filling the first sheet of the table with 10-20 candidates; acceptance based on approved criteria
• If the result is unsatisfactory, the client has the right to stop work without further obligations

Phase 2. Full System — approx. 45%

• All sources, closed platforms manually, evaluation by criteria, filling all table sheets

Phase 3. Handover — approx. 30%

• Instructions, test launch with the client, changing all passwords
• Final payment — after a warranty period of 2–4 weeks, during which the contractor fixes errors for free
• The contractor does not store copies of the code or candidate data

We look forward to collaborating!