About the company: Softermii (www.softermii.com) is an outsourcing/outstaffing company with more than 12 years of experience in web and mobile application development with a team of more than 100 specialists from 6 countries.
We have withstood multiple challenges, such as COVID-19 pandemic and war aggression against Ukraine, we have secured sustainable development processes, saved all our projects and jobs.
Even more — we’re constantly growing. Every day we add new project to our portfolio in the following areas of expertise: * Healthcare * Real estate * Livestream E-commerce * Fintech (WealthTech, Capital Trading) * Real-Time Communications (RTC)
In addition, we develop 2 of our own solutions that help our customers get faster applications for video or audio calls (www.softermii.com/solutions/vidrtc) or an application for real estate with 3D visualization (www.softermii.com/solutions/apartmii)
Company’s mission: Softermii: Succeed Without (L)imitations We will turn any idea into a valuable product.
Our plans as a company: We want to move from a regular outsourcing business to an ecosystem of outsourcing, education, and product development companies. Over the years, we have arrived at a highly personalized working model with transparent workflows. This model allows us to create software products that meet our clients’ goals with precision. We operate in an AI-assisted development environment, where engineering thinking, logical validation, and responsibility for business outcomes are more important than simply writing code.
About the Role: We are seeking a proactive and skilled Middle Python Developer to join our Engineering team and work on — an AI-driven document analysis tool for the legal industry. In this role, you will be directly responsible for building, optimizing, and scaling backend systems that power our next-generation AI agents. You will integrate leading LLM APIs, implement multi-agent workflows, and design robust data pipelines to turn complex AI capabilities into reliable, production-ready software. You’ll join a small, fast-moving team (2 Backend, 1 Frontend, QA, PM) working in a dynamic
Type of cooperation: Remote, Part-time (up to 20 hours per week)
Key Responsibilities: * AI Agent Development: Design, build, and deploy autonomous and semi-autonomous AI agents capable of reasoning, tool use, and task completion * LLM API Integration: Integrate and maintain APIs from providers such as OpenAI, Anthropic, Google Gemini, and open-source models * Orchestration & Frameworks: Implement agentic workflows using modern frameworks * Backend Architecture: Build high-performance, asynchronous REST/gRPC microservices in Python (FastAPI/AIOHTTP) to interface between user-facing services and AI models * Data & Vector Stores: Integrate vector databases (e.g., Qdrant, Pinecone, pgvector) for Retrieval-Augmented Generation (RAG) and memory persistence * Optimization & Monitoring: Implement observability (tracing, logging, evaluation) for LLM interactions to monitor latency, cost, token usage, and response quality * Testing & Robustness: Write unit, integration, and evaluation tests to handle model non-determinism, API rate limits, fallback strategies, and structured output validation
Requirements & Qualifications: * 2-4 years of commercial software development experience with Python * Strong proficiency with FastAPI, Pydantic, and Asyncio for high-concurrency applications * Practical experience with SQL (PostgreSQL) and NoSQL/Redis for session management and caching * Solid understanding of RESTful API design, WebSockets, and event-driven architectures * Comfortable with Docker, Docker Compose, Git, CI/CD pipelines, and basic cloud deployment (AWS/GCP/Azure) * Hands-on experience working with LLM APIs (OpenAI Function Calling/Tools, Anthropic Messages API, etc.) * Understanding of core AI agent patterns (ReAct, Planning, Reflection, Multi-Agent collaboration) * Experience implementing RAG pipelines, chunking strategies, embeddings, and hybrid search with Vector DBs (Qdrant, pgvector, Weaviate, etc.) * Experience with at least one major AI orchestration framework (LangGraph, LangChain, LlamaIndex, CrewAI, or custom native implementations) * Proficiency in enforcing structured data formats from LLMs (via JSON mode, Pydantic Output Parsers, Outlines, or Instructor) * English — В2.
Nice to Have: * Experience with LLM observability & evaluation tools * Familiarity with fine-tuning open-source models * Background in prompt engineering, guardrails (NeMo Guardrails, Guardrails AI), and AI safety/security (prompt injection defense) * Knowledge of graph-based state machines (e.g., LangGraph) for complex agent state management
Why You’ll Love This Role: * State-of-the-art AI work — agentic development is the company’s core driver here, not a side experiment * Real product impact — you’ll build the AI engine behind a legal-tech document analysis tool, not a toy prototype * Broad technical exposure — LLM APIs, agent orchestration, vector search, and async backend architecture all in one role
What Makes This Opportunity Special: * Work at the frontier of agentic AI-driven development as a core company focus, not R&D on the side * Interesting projects across different business stacks * Hands-on experience with modern orchestration frameworks and production LLM systems that are becoming industry-essential * Product-oriented mindset over academic AI experimentation — your work ships and gets used
Who will you have the opportunity to meet during the hiring process (stages): Call, HR, Tech interview, Final interview.
What we can offer you: * We have stable and highly-functioning processes — everyone has their own role and clear responsibilities, so decisions are made quickly and without unnecessary approvals. * You will have enough independence to make decisions that can affect not only the project but also the work of the company. * We are a team of like-minded experts who create interesting products during working hours and enjoy spending free time together. * Do you have a friend who is currently looking for new job opportunities? Recommend them to us and get a bonus. * Do you want to volunteer? We will add you to a chat, where we can get a bulletproof vest, buy a pickup truck or send children’s drawings to the front. * And we have the most empathetic HRs (who also volunteers!). So we are ready to support your well-being in various ways.
A little more information that you may find useful: * Our adaptation period lasts 3 months — this period of time is enough for us to understand each other better. * There is a performance review after each year of our collaboration where we use a skills map to track your growth. * We really have no boundaries in the truest sense of the word — flexible working hours, it’s up to you.
Of course, we have a referral bonus system. So if you have a friend who likes our job opening, you can give us their telegram contact, and we in turn will thank you with a bonus if we start working with them!
