Urgent AI/ML Fix in 48h — RAG Quality, Latency/Cost, Pipeline Stability

Hi, 

I am Olga, Lead AI/ML Engineer based in Warsaw (CET). 
I build and improve production AI systems in Computer Vision (object detection, tracking, face pipelines) and applied LLM/RAG (agents, retrieval, evaluation). My focus is always practical: measurable quality, predictable delivery, and stable operations.

What I will deliver in 48 hours (Basic – $2,500 USDT):

1. Fast technical audit + clear root-cause analysis
2. First working increment shipped (PR / workflow update / patch / prototype)
3. Measurable improvement when possible (quality / latency / cost)
4. Next-milestones plan with estimates and risks

Typical urgent cases I handle

1. RAG quality is bad: irrelevant retrieval, hallucinations, missing context
2. Latency/cost too high: token bloat, inefficient retrieval, missing caching
3. ML pipeline unstable: timeouts, memory spikes, broken ETL, non-reproducible runs
4. CV pipeline issues: detection/tracking instability, poor accuracy, deployment friction
5. “Previous developer left”: need a senior engineer to take over quickly

What I need from you to start today:

1. Repo access / workflow export
2. The #1 blocker + success criteria (what must work after the sprint)
3. Test data or minimal examples
4. Preferred communication channel + availability for a 15-min kickoff

Compute / paid services

To start immediately without delays, you provide your cloud/API credentials (GPU, model APIs, hosting), or we agree on a reimbursable compute deposit upfront. I don’t finance infrastructure out of pocket

Start: same day (CET)
Communication: clear updates, short checkpoints, and deliverables you can verify

Requirements :

1. Access to repo/workflow and sample data
2. Ability to fund the sprint upfront in USDT
3. Client provides cloud/API credentials or reimbursable compute deposit
4. Clear definition of “done” for the sprint (1–2 sentences)