level-3: Sandeep Gunti

[SandeepGunti18]

Submission Level

Final Level 3 submission.

Includes:

• Working AI agent repository
• Uses multiple LPI tools
• Explainable responses
• HOW_I_DID_IT reflection

Repo:
https://github.com/SandeepGunti18/lifeatlas-level3-agent

Signed-off-by: Sandeep Gunti sandeepgunti2004@gmail.com

[github-actions]

✅ Registration + Level 3 received! You're now on the board.

⚠️ Tip for next time: Submit each level as a separate PR — it makes tracking easier and ensures you get credit immediately.

The leaderboard is live: https://life-atlas.github.io/lpi-developer-kit/

Level 3 is what guarantees your place on the team.

[github-actions]

🤖 Level 3 Evaluation

Score: 8/13 ⭐⭐⭐⭐⭐ (62%)

Submission Document

✅ Agent repo linked: https://github.com/SandeepGunti18/ai-career-university-agent
✅ Detailed write-up (191 words)
✅ LPI tools referenced with evidence of actual output
✅ Explainability present — the agent shows its reasoning
✅ HOW_I_DID_IT.md present (124 words)

Repo Analysis

✅ Repo cloned successfully
✅ README present
❌ No evidence of LPI tool calls in code
✅ Error handling present (3 instances)

❓ Your code doesn't appear to call any LPI tools. The requirement is clear: 'Query at least 2 LPI tools.' How does your agent get data from the LPI?

Questions for you

Please respond to these in a comment on this PR. We read every answer.
❓ What choices did you make that weren't in the instructions? What would you do differently next time? We're looking for people who think, not people who follow.

Verdict

Solid foundation. The core is there, but the questions below matter — answer them honestly and push an update.

---

This is an automated evaluation. A human reviewer will follow up on all Level 3 submissions. Scores feed the leaderboard.

Tip: The best way to improve your score is to answer the questions above honestly and push an updated submission.

[github-actions]

✅ Registration + Level 3 received! You're now on the board.

⚠️ Tip for next time: Submit each level as a separate PR — it makes tracking easier and ensures you get credit immediately.

The leaderboard is live: https://life-atlas.github.io/lpi-developer-kit/

Level 3 is what guarantees your place on the team.

[github-actions]

🤖 Level 3 Evaluation

Score: 9/13 ⭐⭐⭐⭐⭐ (69%)

Submission Document

✅ Agent repo linked: https://github.com/SandeepGunti18/lifeatlas-level3-agent
✅ Detailed write-up (191 words)
✅ LPI tools referenced with evidence of actual output
✅ Explainability present — the agent shows its reasoning
✅ HOW_I_DID_IT.md present (124 words)

Repo Analysis

✅ Repo cloned successfully
✅ README present
✅ Code makes real MCP/tool calls (2 references found)
⚠️ Minimal error handling (1 instances)

❓ What happens if the LPI server is down? What if the user sends empty input? What if the LLM returns garbage? Try breaking your own agent and see what happens.

Questions for you

Please respond to these in a comment on this PR. We read every answer.
❓ What choices did you make that weren't in the instructions? What would you do differently next time? We're looking for people who think, not people who follow.

Verdict

Solid foundation. The core is there, but the questions below matter — answer them honestly and push an update.

---

This is an automated evaluation. A human reviewer will follow up on all Level 3 submissions. Scores feed the leaderboard.

Tip: The best way to improve your score is to answer the questions above honestly and push an updated submission.