wiki: sync 2026-09-10 03:12 UTC — 6 file(s) updated
This commit is contained in:
@@ -0,0 +1,88 @@
|
||||
# #aws-jam-2026-chllenge-10
|
||||
|
||||
- **Guild:** `1518726359512387766`
|
||||
- **Channel:** `1547188589484449832`
|
||||
- **Workspace:** `/opt/data/channels/aws-jam-2026-challenge-10`
|
||||
- **Purpose:** Work through AWS Jam 2026 Challenge 10 using the supplied temporary lab account.
|
||||
- **Placement rule:** All channel-owned files and command execution belong in the workspace above.
|
||||
- **Credential rule:** Temporary AWS credentials belong only under the workspace `secrets/` directory and are never copied into this wiki.
|
||||
|
||||
## Challenge tasks
|
||||
|
||||
### Task 10 — Prompt Pandemonium: Restaurant Ruckus
|
||||
|
||||
- **Status:** Chain-of-thought-style operational prompt completed and exercised on 2026-09-09.
|
||||
- **Method:** Numbered parse, reservation-conflict, exact menu-membership, dietary/allergen, prioritization, resolution, and cross-check stages; output exposes only concise evidence and decisions rather than private reasoning.
|
||||
- **Controls:** Per-reservation validation matrix, exact menu-name set membership, exact allergen-set intersection, vegan-label validation, safe escalation, and fixed issue/action schema.
|
||||
- **Model exercise:** Nova Micro followed the required structure and detected the table conflict and vegan mismatch; the prompt was strengthened after the small model showed weak exact-membership reasoning. Titan Premier was authorized but invocation was denied by the lab role.
|
||||
- **Submission prompt:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-010/submission-prompt.txt`
|
||||
- **Evidence:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-010/`
|
||||
|
||||
### Task 9 — Prompt Pandemonium: Flight Fiasco
|
||||
|
||||
- **Status:** Prompt completed and tested on 2026-09-09.
|
||||
- **Method:** Two-example few-shot prompt with a fixed flight-confirmation schema and fail-safe handling for missing, conflicting, corrupted, and additional values.
|
||||
- **Model test:** Amazon Nova Micro in `us-east-1` was `AUTHORIZED`; invocation succeeded.
|
||||
- **Verification:** The generated confirmation preserved every supplied passenger and flight detail, maintained the exact field order, and retained an extra meal preference under `Additional details`.
|
||||
- **Submission prompt:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-009/submission-prompt.md`
|
||||
- **Evidence:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-009/`
|
||||
|
||||
### Task 8 — Prompt Pandemonium: Booking Blunder
|
||||
|
||||
- **Status:** Prompt completed and tested on 2026-09-09.
|
||||
- **Method:** COSTAR prompt with explicit Context, Objective, Style, Tone, Audience, and Response requirements.
|
||||
- **Model test:** Amazon Nova Micro in `us-east-1` was `AUTHORIZED` and `AVAILABLE`; invocation succeeded.
|
||||
- **Verification:** Sample output included the customer name, dates, room type, total cost and currency, complete cancellation terms, and all hotel contact methods without inventing values.
|
||||
- **Submission prompt:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-008/submission-prompt.md`
|
||||
- **Evidence:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-008/`
|
||||
|
||||
### Task 7 — The Lord Of The RAGs, Task 4
|
||||
|
||||
- **Status:** Completed and verified on 2026-09-09.
|
||||
- **Translation:** Amazon Translate converted the English data-center query to German (`de`).
|
||||
- **RAG result:** Nova Lite returned a German response describing three Tier IV data centers across the United States, Europe, and Asia, with 150,000 square feet of combined raised-floor space and resilient cooling, power, and networking.
|
||||
- **Submission:** The translated response was entered in the required widget; the unchanged Task 4 cell completed successfully; `task4.json` exists and contains a 256-value Titan embedding. No later code cells were executed.
|
||||
- **Evidence:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-007/`
|
||||
|
||||
### Task 6 — The Lord Of The RAGs, Task 3
|
||||
|
||||
- **Status:** Completed and verified on 2026-09-09.
|
||||
- **Knowledge Base sync:** First ingestion job reached `COMPLETE`; 18 new documents were indexed and 2 documents failed processing.
|
||||
- **Prompt:** Requested an English summary of Octank Financial Corp.'s data-center locations, capacity, infrastructure, reliability, and operations.
|
||||
- **RAG result:** Returned a grounded English response describing three Tier IV data centers across the United States, Europe, and Asia with 150,000 square feet of combined raised-floor space and resilient cooling, power, and networking.
|
||||
- **Submission:** The unchanged Task 3 cell completed successfully; `task3.json` exists and contains a 256-value Titan embedding. No later code cells were executed.
|
||||
- **Evidence:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-006/`
|
||||
|
||||
### Task 5 — The Lord Of The RAGs, Task 2
|
||||
|
||||
- **Status:** Completed and verified on 2026-09-09.
|
||||
- **Knowledge Base:** `aws-KB-jam-317783770379`, ID `Z10DYJHIPA`, discovered as the single active Knowledge Base in `us-east-1`.
|
||||
- **Notebook:** Rendered the Knowledge Base widget, set its value to the discovered ID, and executed through the unchanged `submit_your_answer('task2.json', kb_id.value)` cell.
|
||||
- **Verification:** Submission cell returned `ok`; generated `task2.json` contains the exact Knowledge Base ID; no code cells after Task 2 were executed.
|
||||
- **Evidence:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-005/`
|
||||
|
||||
### Task 4 — The Lord Of The RAGs, Task 1
|
||||
|
||||
- **Status:** Completed and verified on 2026-09-09.
|
||||
- **Region:** `us-east-1`; SageMaker notebook instance `SageMakerNotebookInstance-zCn04RrlkgD6`.
|
||||
- **Models:** Amazon Nova Lite and Titan Text Embeddings V2 both reported `AUTHORIZED` and `AVAILABLE`; both notebook invocation tests passed.
|
||||
- **Notebook:** Executed code cells `6`, `8`, `10`, `12`, and `14`, ending with `submit_your_answer('task1.json', 'task1')`; no later code cells were executed.
|
||||
- **Submission:** `task1.json` was listed in the notebook submission bucket with last-modified time `2026-09-09T10:47:28Z`.
|
||||
- **Evidence:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-004/`
|
||||
|
||||
### Task 3 — Validate latest Rekognition category
|
||||
|
||||
- **Status:** Completed and verified on 2026-09-09.
|
||||
- **Test:** Uploaded a fresh JPEG object and matched its exact S3 key to the resulting DynamoDB `ProductId`.
|
||||
- **Result:** DynamoDB item count increased from `1` to `2`; the new item category is `Home Decor`.
|
||||
- **CloudWatch constraint:** The lab session policy denies `logs:FilterLogEvents`; successful downstream DynamoDB creation proves the Lambda pipeline processed the upload.
|
||||
- **Evidence:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-003/`
|
||||
|
||||
### Task 2 — Lambda processing window
|
||||
|
||||
- **Status:** Completed and verified on 2026-09-09.
|
||||
- **Function:** `product-categorization-function` in `eu-west-2`.
|
||||
- **Change:** Timeout increased from `3` to the Lambda maximum of `900` seconds.
|
||||
- **Verification:** Function is `Active`; `LastUpdateStatus` is `Successful`; code hash and execution role are unchanged.
|
||||
- **CloudWatch constraint:** The lab session policy denies `logs:FilterLogEvents`; configuration was verified through the permitted Lambda API.
|
||||
- **Evidence:** `/opt/data/channels/aws-jam-2026-challenge-10/challenges/task-002/`
|
||||
Reference in New Issue
Block a user