Week #3 #
Implemented MVP features #
User login and registration, PR
Name of the profile, PR
Picture in the profile, PR
Possibility to rotate the map, (was implemented on previous week)
Possibility to open labels with fish pictures on the map, PR
Adding fishing places, PR back
Possibility to start the event of fishing, PR Front, PR Back
Test model for ML, PR
Demonstration of the working MVP #
Link to the video: Google drive
ML #
Remove this section if not applicable
Link to dataset: Link
We build UML diagram for the ML part and started implementing some parts of it:
- Vector DB: Based on Quadrant cloud solution, to store ready embeddings
- Fish Speices class: to store all information about row in one place
- Data preparation: to prepare data for tokenization in our embedder
Links to the initial model artifacts: Qwen3-Embedding-0.6B
Internal demo #
All needed comments were provided on the video.
Weekly commitments #
Individual contribution of each participant #
Stepan Vagin:
- Created UML of the architecture, Link
- Created classes FishSpeices, VectorDatabase, to connect to vector DB, PR
- Created first ML endpoint, PR
- Updated backlog, Link
Kostya Zimin:
- fix db migrations and add liquibase, PR
- refactor all controllers for useful explanation in swagger and create API contract, PR
- fix all methods in service part and add more exceptions, PR
- fix sql codes, db relationships and docker-compose file for back part
Ivan Vasiliev:
- defined the architecture of search system with Stepan Vagin(discussed which models, dataset to use, workflow)
- developed preprocess of the data, added embeddings of the data PR retrieved datasets from the data_prep.py, Link
- worked on localization of the dataset, compared different models(Helsinki-NLP/opus-mt-en-ru, facebook/nllb-200-distilled-600M, mistral-7b-instruct-v0.1.Q4_K_M.gguf) mess work can be found here, Link
Kirill Greshnov:
- Connect login and registration with backend, add page for fishing event with statistics, PR
- Add local image upload for fishing event, PR
- Add feature to choose fishing location, PR
- Add functionality to profile page to show actual user data, minor design changes, PR
Egor Belozerov:
- containerize the ML part + add it to deployment, PR
- additional build and linting flutter app, PR
- fix 301 status code for https endpoint, PR
- deploy the backend, PR
- fix the issues with runner and sops (recreate the encryption keys)
Damir Sadykov:
- Started working on final app design (some screens are done, some sketches for other were made, color palette with fonts has been selected) in Figma: Link
- Research in automated tools for design travel from Figma to Flutter (plugins: “Figma to Flutter”, “Figma2Flutter”, “Supernova”, tools: “FlutterFlow”). Result: no plugins are good enough for this purpose (lot’s of bugs) and tool FlutterFlow is good for transfer design to flutter code.
- FlutterFlow design for first screen are made, FlutterFLow (subscription required, so screenshots are made (in same Figma file: Figma)
- Fishing event related requests handle are made: fishing event is now ready, PR
- Pictures of catches upload request handled too, PR
Adel Gaznanov:
- Сreate endpoint to add new water places on the map, PR
- Сreate endpoints to start and end fishing on the specific water place, PR
- Сreate an opportunity to save caught fish during fishing, PR
Plan for Next Week #
- Start to implement features, meant for final project
- Continue building ML architecture
- Fixing bugs
Confirmation of the code’s operability #
We confirm that the code in the main branch:
- [+] In working condition.
- [+] Run via docker-compose (or another alternative described in the
README.md
).