Machine Learning Engineer

Were building a Multimodal AI Companion Platform (Telegram Bot + Web Dashboard) where creators can monetize personalized AI clones through text, voice, image, and video interactions.

Were looking for a Senior Machine Learning Engineer to own model fine-tuning, personalization pipelines, and production-grade ML delivery, working closely with backend systems under strict timelines.

This is not a research role. This is about shipping real ML into production, fast.

The Mission:

Phase 1 Build (MVP Delivery)

Fine-tune and adapt LLMs to capture creator personality and conversational style

Implement RAG pipelines that feel consistent, contextual, and in-character

Ship stable ML inference flows that integrate cleanly with backend APIs

Phase 2 Launch (Pilot Rollout)

Optimize latency, cost, and reliability for real user traffic

Ensure voice, image, and text generation pipelines are production-ready

Actively debug and refine models based on real interaction data

Phase 3 Scale (Quality + Cost Control)

Improve personalization quality without blowing up GPU costs

Help shape longer-term ML architecture (LoRAs, memory, inference patterns)

Support roadmap decisions with realistic ML constraints and tradeoffs

Core Responsibilities:

Model Fine-Tuning & Personalization

Fine-tune LLMs for conversational style, tone, and behavioral consistency

Train and manage LoRA adapters for image/video personalization

Continuously improve response quality using real interaction feedback

RAG & Memory Systems

Design and maintain RAG pipelines using vector databases

Balance retrieval accuracy, latency, and token cost

Ensure long-term memory feels natural and coherent across sessions

Multimodal ML Pipelines

Own voice cloning and TTS workflows (training + inference)

Integrate STT, TTS, image, and video generation into a unified experience

Ensure outputs are reliable, high-quality, and monetization-ready

Production ML & Backend Integration

Build ML systems that work cleanly with FastAPI-based backends

Handle async workflows, background jobs, and inference queues

Optimize for serverless GPU execution and cost efficiency

Delivery Under Tight Timelines

Ship working ML features within aggressive deadlines

Make pragmatic tradeoffs between perfect and production-ready

Take ownership when things break and fix them fast

Requirements:

Strong Applied ML Experience

5+ years in Machine Learning / Applied AI

Hands-on experience with LLM fine-tuning, embeddings, and RAG

Experience deploying ML systems used by real users

Backend-Aware Engineer

Strong Python skills beyond notebooks

Comfortable working with APIs, async flows, and production constraints

Experience collaborating closely with backend engineers

Multimodal Exposure (Preferred)

Voice cloning / TTS / STT pipelines

Image or video generation with diffusion models

Serverless GPU inference platforms

Execution Mindset

You prioritize shipping over theory

Youre comfortable with ambiguity and fast decisions

You can work independently without heavy process overhead

Why Join

Real Ownership: You own ML quality, not just experiments

High Impact: Models you ship directly drive revenue

Speed: Small team, aggressive timelines, zero bureaucracy

Depth: Work across LLMs, RAG, voice, image, and infra not just one slice