You will get your first AI task set up with the tools you already have
Rising Talent
Rising Talent
Project details
You already pay for ChatGPT, Claude, or Copilot. This project turns that subscription into one working task for your business — set up, tested on your real examples, and documented so your team can run it without me.
Pick a task like: drafting replies to common customer emails, summarizing meeting notes into action lists, turning inquiries into quote-ready summaries, first-draft job posts, product descriptions, or weekly report drafts. Not sure which? Tell me what eats your time and I'll suggest the highest-value first task.
What you get: a reusable setup — saved prompt, Custom GPT, or Claude Project — built and checked against 2–3 of your real examples, a step-by-step playbook with screenshots your team can follow, and troubleshooting for the most likely failure points. Starter covers one task; Standard covers three, each with its own playbook, plus a captioned walkthrough video you can rewatch anytime. Prefer to talk it through live? A 30-minute call is an optional add-on — the documentation stands on its own without it.
No code, no API keys, no new software to buy: this works inside the subscription you already have.
Pick a task like: drafting replies to common customer emails, summarizing meeting notes into action lists, turning inquiries into quote-ready summaries, first-draft job posts, product descriptions, or weekly report drafts. Not sure which? Tell me what eats your time and I'll suggest the highest-value first task.
What you get: a reusable setup — saved prompt, Custom GPT, or Claude Project — built and checked against 2–3 of your real examples, a step-by-step playbook with screenshots your team can follow, and troubleshooting for the most likely failure points. Starter covers one task; Standard covers three, each with its own playbook, plus a captioned walkthrough video you can rewatch anytime. Prefer to talk it through live? A 30-minute call is an optional add-on — the documentation stands on its own without it.
No code, no API keys, no new software to buy: this works inside the subscription you already have.
AI Algorithms
Large Language ModelAI Applications
AI Content Creation, Natural Language GenerationAI Tools
Microsoft 365 CopilotAI Models
ChatGPTWhat's included
| Service Tiers |
Starter
$150
|
Standard
$295
|
Advanced
$495
|
|---|---|---|---|
| Delivery Time | 5 days | 7 days | 10 days |
Number of Revisions | 1 | 1 | 2 |
AI Model Integration | - | - | - |
Batch Normalization | - | - | - |
Database Integration | - | - | - |
Detailed Code Comments | - | - | - |
Image Upscaling | - | - | - |
MLOps | - | - | - |
Model Deployment | - | - | - |
Model Documentation | - | - | - |
Model Monitoring | - | - | - |
Model Testing & Optimization | - | - | - |
Model Tuning | - | - | - |
Natural Language Processing | - | - | - |
NLP Tokenization | - | - | - |
Pre-Training | - | - | - |
Prompt Engineering | |||
Setup File | - | - | - |
Source Code | - | - | - |
Optional add-ons
You can add these on the next page.
Fast Delivery
+$50 - $75
30-minute live walkthrough call
(+ 2 Days)
+$150
One additional task with its own playbook
(+ 2 Days)
+$95Frequently asked questions
About Joanna
AI/ML Engineer | LLM Evaluation, RAG Accuracy & Reliability | Python
Naperville, United States - 4:37 am local time
I work as an AI engineer in medical AI, where I build and rigorously evaluate production LLM pipelines: automated evaluation of output quality (eval harnesses, gold test sets, LLM-as-judge reliability checks, regression tests for nondeterministic outputs), retrieval systems, and API resilience layers (error classification, retry, circuit breaking, cost tracking). My focus is the part that makes AI dependable in production, not just impressive in a demo.
What I do for clients:
- RAG / chatbot accuracy audits — test suites built from your documents, scored for accuracy and faithfulness, with every failure traced to its root cause: retrieval, prompt, or model (see my project catalog)
- LLM evaluation pipelines — golden datasets, quality metrics with thresholds as CI gates, judge/scorer reliability testing
- Production LLM engineering — Claude, OpenAI, and Azure OpenAI APIs: structured outputs, tool use, prompt engineering, FastAPI + Docker services with tests and CI
- AI adoption for small businesses — workflow audits and safe, cost-controlled implementation, with the same testing discipline
Representative work (public on GitHub, linked in my portfolio): an agentic clinical-reasoning system for a Harvard drug-decision benchmark (21st of 323 teams, 2,491 questions, live medical APIs); a production-style FastAPI + LLM service with tests, CI, and Docker; a Claude-API document-automation pipeline; and a real-time meeting assistant (Deepgram + Claude with prompt caching, ~90% session cost reduction).
Academically: M.S. in Artificial Intelligence & Machine Learning, B.S. in Computer Science, and an in-progress M.S. in Biotechnology/Bioinformatics for extra depth in biomedical and scientific data.
Clients can expect clear communication, tested and well-structured code, and honest assessments — including when the right answer is "don't automate this." If you need someone who can tell you whether your AI can be trusted, and then make it trustworthy, I'd be glad to help.
Steps for completing your project
After purchasing the project, send requirements so Joanna can start the project.
Delivery time starts when Joanna receives requirements from you.
Joanna works on your project following the steps below.
Revisions may occur after the delivery date.
Task selection and scope
I review your answers and confirm the task — or recommend one based on what eats your time — with what's in and out of scope, before any work starts.
Build and check against your real examples
I build the prompt setup in your AI tool and check it against your real examples, adjusting until the outputs are usable first drafts.