Mightyzeus

Software engineer

[Mightyzeus @mightyzeus9]

Actively looking for a Go lang role

[Mightyzeus @mightyzeus9]

Introducing Doc explain

Doc Explain is a RAG-as-a-Service backend(Go lang) that lets users upload documents, processes them asynchronously, and enables document-aware Q&A with persistent chat context.

Take a look at https://t.co/pc7nkJiiTL
#golang #backend #ai #aiengineering

[Mightyzeus @mightyzeus9]

I think I have built this RAG system to a point where I can show it, will deploy it and share

Interesting times for real🙌🏿🙌🏿

[Mightyzeus @mightyzeus9]

I think I have built this RAG system to a point where I can show it, will deploy it and share

Interesting times for real🙌🏿🙌🏿

[Mightyzeus @mightyzeus9]

Since I integrated FastLane into all the projects I led it work on

Life has been easier releasing builds

[Mightyzeus @mightyzeus9]

document chunks are now encrypted before being stored in the database and decrypted only when needed for retrieval/LLM responses, reducing exposure of raw document text.

[Mightyzeus @mightyzeus9]

Update on RaaS:
I’ve added document-level conversation persistence, so users can ask natural follow-up questions without repeating context. The system remembers recent chat history while still grounding every answer in the document.

[Mightyzeus @mightyzeus9]

Update on RaaS:
I’ve added document-level conversation persistence, so users can ask natural follow-up questions without repeating context. The system remembers recent chat history while still grounding every answer in the document.

[Mightyzeus @mightyzeus9]

Update on RaaS:
I’ve added document-level conversation persistence, so users can ask natural follow-up questions without repeating context. The system remembers recent chat history while still grounding every answer in the document.

[Mightyzeus @mightyzeus9]

added core instructions so i don't get hallucinated result when retriving context

[Mightyzeus @mightyzeus9]

Raggo Embedding service, still experimenting tho, chunk size of 250 seems to be big, when i implement the retrival flow, it would hlep me decide better if answers are accurate

[Mightyzeus @mightyzeus9]

cause raggo chokes on some documents while parsing, openai is used as a fallback and neccessary update is made to the DB

[Mightyzeus @mightyzeus9]

code snippet using openai go ask to extract parse document to get it ready for chunking using raggo #golang #backend #engineering #ai

[Mightyzeus @mightyzeus9]

Using raggo built in document parser but it’s choking a even pdf files, now all pdf files are being parsed successfully. Since I use open ai models for embeddings, I would implement parser with their SDK. Make sense to just use one provider #golang #openai #ai

[Mightyzeus @mightyzeus9]

Using raggo built in document parser but it’s choking a even pdf files, now all pdf files are being parsed successfully. Since I use open ai models for embeddings, I would implement parser with their SDK. Make sense to just use one provider #golang #openai #ai

[Mightyzeus @mightyzeus9]

@webdesignerng Mental health

[Mightyzeus @mightyzeus9]

* Been working on a..

[Mightyzeus @mightyzeus9]

Been working RAG-powered Housing-Agent for Lagos tenancy questions.
It’s based on the Lagos Tenancy Law, 2011, and answers user questions with supporting excerpts from the source document, so you get clarity plus receipts.

Still in the kitchen cooking tho🧑‍🍳
#golang #go #ai

[Mightyzeus @mightyzeus9]

this is a go lang powered system btw

[Mightyzeus @mightyzeus9]

Been working RAG-powered Housing-Agent for Lagos tenancy questions.
It’s based on the Lagos Tenancy Law, 2011, and answers user questions with supporting excerpts from the source document, so you get clarity plus receipts.

Still in the kitchen cooking tho🧑‍🍳
#golang #go #ai

[Mightyzeus @mightyzeus9]

@abolorreeeee It won’t be a general thing or wrapper, say you have two endpoints, one might need it, one won’t. Scenarios where you might need it would be like if you have to update multiple records at the same time, user a worker pool of like 10 and retry logic and should be good.

[Mightyzeus @mightyzeus9]

Before the hybrid search, the distance I was getting for the answer were between 0.78-0.93, after implementing, same question, I am now getting between 0.3 -0.5

[Mightyzeus @mightyzeus9]

Implemented a Hybrid search in a RAG system I am working and the answer seems more accurate and faster. Will study this and see if a I need a re ranker or not

[Mightyzeus @mightyzeus9]

Implemented a Hybrid search in a RAG system I am working and the answer seems more accurate and faster. Will study this and see if a I need a re ranker or not

[Fathiu @abolorreeeee]

Completed the Takeout Order refund system for my restaurant management system in Golang

Here's what makes it production grade:

𝗜𝗱𝗲𝗺𝗽𝗼𝘁𝗲𝗻𝗰𝘆
→ check refund table before calling Paystack
→ same order refunded twice = rejected immediately
→ uniqueIndex on order_id enforces this at DB level

𝗔𝘁𝗼𝗺𝗶𝗰 𝗧𝗿𝗮𝗻𝘀𝗮𝗰𝘁𝗶𝗼𝗻
→ refund record created
→ order status → cancelled
→ payment status → refunded
→ outbox event written
→ all in ONE transaction, all succeed or all fail

𝗢𝘂𝘁𝗯𝗼𝘅 𝗣𝗮𝘁𝘁𝗲𝗿𝗻
→ refund email event written to outbox inside transaction
→ if Redis is down → outbox worker retries
→ customer always gets refund notification
→ no event lost

𝗔𝘂𝗱𝗶𝘁 𝗧𝗿𝗮𝗶𝗹
→ dedicated refunds table
→ own idempotency key
→ reason stored
→ processed_at timestamp
→ every refund traceable forever

Stack: Go, PostgreSQL, Paystack, Redis Streams

GitHub → https://t.co/aY2N7Rc2dW

Working on the Table reservation refund system

[Mightyzeus @mightyzeus9]

@abolorreeeee You be senior man nah🙌🏿🙌🏿