2
2 Comments

Sold-as-is: AI nutrition tracker (4 months solo dev, 47K LOC, Google Play approved, $25K)

TL;DR: I built a production-ready AI nutrition tracker over 4 months (47K LOC, Google Play approved, 12 languages, Architecture C cascade pipeline with GPT-5.4-mini + Claude). Selling for $25K (76% below the $105K+ replacement cost). Pre-revenue but launch-ready. Listed on Little Exits.

The story

I'm Billel, solo dev from France. Over the last 4 months I've been building Calpyra AI - an AI nutrition tracker that scans any meal photo and returns calories + macros + nutrition in 12 languages, in under 3 seconds.

I have 3 other live projects competing for my time (Cal-Halal, Minutiva AI) and Calpyra hit "production-ready" before "marketed". Rather than letting it stagnate, I'd rather see it succeed in operator hands.

What's actually in the box

Codebase metrics:

  • 47,165 lines of production code (40K TypeScript frontend + 7K Python backend)
  • 293 commits YTD 2026 (4 months solo dev)
  • 21 polished screens, 70 React components, 29 Zustand stores
  • 24 backend FastAPI endpoints, 142 pytest tests passing
  • 12 locale JSON files, 9,600+ translation strings (RTL ready)

AI pipeline (Architecture C):

  • Tier 1: GPT-5.4-mini Vision single-pass (88% of scans @ $0.001 each)
  • Tier 2: self-consistency N=3 + median merge for ambiguous photos
  • Tier 3: GPT + Claude Sonnet 4.5 cross-model ensemble for low-confidence
  • Custom-trained YOLOv12 food-detection model (proprietary asset)
  • Depth Anything v2 Small ONNX for portion-volume estimation

Distribution:

  • Google Play Internal Testing: AAB v1.1.0 (versionCode 20) already approved
  • Health Connect Android integration validated on Xiaomi/Pixel
  • iOS HealthKit wired (untested binary, ~1-2 weeks to ship)
  • Backend live on Railway
  • Public landing page (8-screen workflow demo): https://billel-abbas.github.io/calpyra-acquire/

Why $25K

Documented replacement cost: $105K+ (line-by-line in the CIM):

  • Mobile app dev: $45K (4 mo solo dev @ $80/h)
  • Backend: $15K
  • YOLOv12 fine-tuning: $8K
  • Architecture C cascade: $12K
  • Health Connect integration: $5K
  • 12-language i18n (9,600 strings @ $0.50): $4.8K
  • ...plus brand, tests, Play Console setup, etc.

At $25K = 76% discount vs. building from scratch.

It's also at the Little Exits Premium cap for pre-revenue listings.

What you'd inherit

  • Full GitHub repo ownership (293 commits transferred to your org)
  • Google Play Console (Calpyra Studio, $0 transfer fee)
  • Railway backend project + Firebase project
  • EAS Build pipeline with service account JSON
  • Custom YOLOv12 model weights + training pipeline scripts
  • 12-language locale files (full IP)
  • "Calpyra" brand (trademark-free on USPTO + EUIPO)
  • 7-day post-acquisition support

5 personal guarantees

  1. 30-day "Build Working" guarantee - full refund if you can't get it running locally
  2. Code originality - 100% original or MIT/Apache deps, no GPL contamination
  3. Play Store transfer guarantee
  4. No hidden costs guarantee - all third-party services itemized
  5. Source code completeness guarantee

Where to look

Who wins

  • Indie hackers wanting to skip 4-6 months of dev
  • Wellness brands / nutrition coaches with audience but no app
  • AI/ML engineers wanting a reference cascade pipeline + custom-trained model
  • Existing nutrition app owners (rebrand + expand to 12 languages instantly)

Not for: passive investors expecting cash-flow on day 1. MRR is $0 today. Asset is launch-ready, not launched.

Reach out

Drop a comment, DM me here, or message via the Little Exits listing. Happy to do a 30-min technical walkthrough call post-NDA (recorded).

PS - If anyone has feedback on the listing description or the landing page, I'm all ears. First time selling a project, so iterating live.

on May 18, 2026
  1. 1

    This is a strong asset, but I think the listing is leaning too much on replacement cost and codebase depth. For a buyer, the sharper story is probably “launch-ready AI nutrition app with proven technical risk already removed.” The 47K LOC, Play approval, Health Connect, 12 languages, cascade AI pipeline, and custom YOLO model all matter because they reduce build risk.

    The biggest gap is market framing. “AI nutrition tracker” is a crowded category, but Calpyra could be positioned more specifically around instant meal intelligence for wellness brands, coaches, or regional nutrition audiences that do not want to build the app layer from scratch.

    One thing I’d watch is the name. Calpyra is decent, but if a buyer wants to relaunch this as a softer wellness/health brand, Lyriso.com would probably fit the consumer nutrition direction better and make the asset feel less like a technical build-for-sale project.

  2. 1

    This is a seriously impressive engineering-heavy build — especially the cascade inference pipeline + multi-model setup. You can tell this was built by someone optimizing for real-world accuracy, not just a demo.

    A few things that stand out immediately:
    • Multi-tier AI architecture (vision → consistency → ensemble) is a strong approach for noisy nutrition data
    • YOLOv12 + depth estimation for portion sizing is not trivial — that’s real ML work, not wrapper AI
    • 12-language + RTL support from day one shows solid production thinking
    • Health Connect + Play Store approval adds real distribution value beyond code

    On the positioning side, the hardest part here isn’t the code — it’s the fact that it’s pre-revenue. Buyers for technical assets like this usually want at least one of:
    • active user traction
    • existing revenue
    • or a very clear distribution channel

    Without that, the value becomes mostly:
    • speed-to-market advantage
    • and reusable ML + app architecture

    One suggestion that could strengthen the listing:
    show a “first 100 users” acquisition plan or wedge (fitness creators, calorie tracking niche, TikTok demo loops, etc.). That helps buyers see how it becomes revenue, not just how it was built.

    Also worth emphasizing even more:
    the cascade system + dataset pipeline is probably the real IP here, not the app UI.

    Emmanuel here 👋
    If you’re open to feedback on the technical architecture, GTM strategy, or packaging this for acquisition buyers, I’d be happy to take a closer look or collaborate on refining it.

    Calpyra Acquisition Listing

    https://teams.live.com/l/invite/FAAk3iOSJkDyS11JQE?v=g1