I let open-source AI name the birds, so my phone could stay in my pocket

Snehal Thombare — Sun, 11 Oct 2026 15:41:50 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

What I Built

Trail Journal is an offline bird-walk companion. You record a walk with your phone in your pocket, then a local pipeline tells you which birds it heard and writes a short journal entry about the walk in Marathi. The screen is the shortest part of the experience: the walk is the point.

How it works

BirdNET-Analyzer (open-source, runs offline) listens to the recording and lists the birds with confidence scores.
Gemma 3 4B, an open-weight model running locally through Ollama, turns that list into a Marathi journal entry.
A ~40-line Python script glues them together and writes journal.md.

Test run on an Asian Koel recording from xeno-canto: 27 detections, average confidence 0.78.

Output of journal.md: bird list from BirdNET, then the Marathi entry from Gemma 3 4B. Bird and place names stay in English, the sentences are Marathi.

What went wrong (and how I fixed it)

My first prompt made Gemma mention "other birds" that BirdNET never detected, and the Marathi had a stray Hindi word and some garbage text. A stricter prompt fixed it: write only about the data, never invent a bird, use simple correct Marathi.

Small local models need tight guardrails, but with them they do the job.

Why open matters here

No signal needed. Record on the trail, process at home. After one-time model downloads (BirdNET ~224 MB, Gemma ~3 GB), nothing needs the internet.
My recordings stay on my laptop. No audio goes to a server I don't control.
Free. No API keys or credits.
Swappable. Changing the model or prompt is a one-line edit, which is how I fixed the hallucination.
Marathi. I wanted the journal in my own language.

Limitations

My first test used a xeno-canto recording, not my own walk.
Gemma 3 4B's Marathi is good but not perfect, so I check it before sharing.
BirdNET confidence is a guide, not proof.

Code

https://github.com/binarymind-dev/Trail-journal

Credits: BirdNET-Analyzer (Cornell Lab of Ornithology and Chemnitz University of Technology, non-commercial license); test audio XC1173496 by Lars Lachmann via xeno-canto; Gemma by Google via Ollama.

How I’m Learning C++ and DSA as a CSE Student 🚀

Snehal Thombare — Fri, 02 Oct 2026 00:33:39 +0000

How I’m Learning C++ and DSA as a CSE Student 🚀

As a Computer Science student, one of the areas I’ve been focusing on is problem solving through C++ and Data Structures & Algorithms (DSA).

When I started competitive programming, I realized that learning a programming language is only the beginning. The real challenge is learning how to think about a problem, find an efficient approach, and implement it correctly.

Why I Chose C++

I decided to focus on C++ because it gives me a strong foundation for problem solving and competitive programming.

Some of the things I’m currently working with are:

C++ fundamentals
STL
Arrays and Strings
Linked Lists
Stacks and Queues
Sorting and Searching
Recursion
Greedy algorithms
Basic problem-solving patterns

The STL has been especially useful because containers such as vector, set, map, and unordered_map make it easier to focus on the actual problem rather than implementing everything from scratch.

My Approach to DSA

Instead of only watching tutorials, I’m trying to learn DSA by actually solving problems.

My current process is:

Understand the problem carefully.
Think about a simple approach.
Check its time and space complexity.
Look for a more efficient solution.
Implement it in C++.
Test it with different cases.
Review the solution and understand what I could improve.

This process has helped me understand that DSA is not about memorizing solutions. It is about recognizing patterns and developing a systematic way of thinking.

Competitive Programming

I’m also exploring competitive programming through platforms such as LeetCode and Codeforces.

Competitive programming has taught me some important lessons:

Read the constraints before solving.
Think about complexity before coding.
Test edge cases.
Don't give up when the first approach fails.
Learn from wrong submissions.
Practice consistently.

Some problems that look simple at first can require a completely different approach once the constraints become large.

What I’m Building Alongside DSA

I don’t want my learning to be limited to solving problems.

Alongside DSA, I’m also exploring:

Backend development
AI/ML
Full-stack development
Open source
AI-powered applications

Building projects helps me connect programming concepts with real-world applications.

For me, the combination of DSA + development + projects is helping me understand both problem solving and software development.

What I’m Learning From This Journey

The biggest lesson so far is that consistency matters more than trying to learn everything at once.

There are days when a problem takes only a few minutes, and there are days when I spend much longer trying to understand a solution.

Both are part of the learning process.

I’m still learning, experimenting, and improving one problem at a time.

What's Next?

My next goals are to:

Improve my DSA skills
Solve more medium-level problems
Participate in more competitive programming contests
Build stronger backend and AI projects
Contribute to open source
Document what I learn along the way

I’m excited to continue this journey and share more of my learning here on DEV.

If you're also learning C++, DSA, competitive programming, AI, or software development, feel free to connect and share what you're working on! 👋

One problem. One project. One lesson at a time. 🚀

DEV Community: Snehal Thombare

I let open-source AI name the birds, so my phone could stay in my pocket

What I Built

How it works

What went wrong (and how I fixed it)

Why open matters here

Limitations

Code

How I’m Learning C++ and DSA as a CSE Student 🚀

How I’m Learning C++ and DSA as a CSE Student 🚀

Why I Chose C++

My Approach to DSA

Competitive Programming

What I’m Building Alongside DSA

What I’m Learning From This Journey

What's Next?