About the Role

We are looking for excellent Software Engineers who can design challenging software engineering tasks to train AI models. Each task should require a minimum of 1.5 hours -2 hours of screen recording to complete and must use VS Code along with at least one of the approved tools listed below. After proposing the task, and getting the approval on the same, you will complete it while recording your screen and explaining your thought process throughout the implementation.

Guidelines can be checked here -

- v5-expert-tooling-recording-guidelines
- Expert Guide: Creating Long-Horizon Tasks

Please read the complete and guidelines before starting the assessment: https://forms.gle/fNMQeF2F5Xa3DaZ59

Complete this assessment within 24 hours . As we are hiring on an urgent basis, failure to submit within 24 hours may result in your candidature being cancelled.

Responsibilities

- Propose complex software engineering tasks that require at least 1.5 hours to complete.
- Complete the proposed task while recording your screen and explaining every step.
- Use VS Code along with at least one of the approved tools listed below.
- Deliver a minimum of 6 hours of approved screen recordings per day .
- Spend approximately 2–3 hours per day creating high-quality task proposals.
- Deliver 150+ hours of approved recordings per month .
- Follow all project guidelines and quality standards.

Approved Tools Creative/media: Blender, GIMP 2.10 + GIMP 3.2.4 (gimp3), Krita, Inkscape, Kdenlive, OBS Studio, Audacity, Scribus, Synfig Studio, Pencil2D, LMMS, Ardour

CAD / 3D / Sci: FreeCAD, LibreCAD, OpenSCAD, ParaView, GeoGebra, Logisim, SolveSpace

3D / Game / VFX: Godot 4.6.1, Natron 2.5.0

Dev / IDE / Diagram: VS Code, IntelliJ IDEA 2025.3, PyCharm 2025.3, yEd 3.25.1, Wireshark, Graphviz, Node-RED

Office: LibreOffice

Agent Scripting (Python): ezdxf, numpy-stl, trimesh, opencv, librosa, mido, pretty_midi, soundfile, svgelements, pypdf, pdfplumber

Eligibility

- Solid software engineering and problem-solving skills.
- Excellent understanding of programming and system design.
- Comfortable explaining technical concepts while recording the screen.




- Immediate joiners preferred.

Working Hours

- Monday to Saturday
- 9 Hours per Day

Selection Process

- Project guideline review
- One technical assessment
- Final selection based on assessment performance

Example Task Proposal Task 1: Configurable Async Rate Limiter with Token Bucket, Circuit Breaker, and Observability

Implement a production-style asynchronous rate-limiting framework for an HTTP client that enforces configurable per-host concurrency and request-rate limits while remaining resilient to transient service failures. The implementation must be driven entirely by a provided configuration specification, support retries with exponential backoff and jitter, expose structured metrics and logs, and include a comprehensive automated test suite validating concurrency limits, retry timing, circuit-breaker behavior, and overall throughput.

Workflow:

A configuration specification document describing per-host limits (maximum concurrent requests, requests-per-second, retry policy, exponential backoff formula with jitter, and circuit-breaker thresholds).

A high-level architecture/design note describing the intended request processing pipeline, component responsibilities, and interactions between the scheduler, token bucket, retry handler, circuit breaker, logging, and metrics modules.

A mock dataset of 20-50 API endpoints spanning multiple hosts (e.g., api-a.local, api-b.local, api-c.local).

A configurable fakeRequest(url) implementation that simulates:

- Variable latency (100-500 ms)
- Random transient failures (~20%)
- Host-specific failure patterns
- Configurable timeout scenarios

An intentionally incomplete HTTP client codebase that currently performs unrestricted asynchronous requests without rate limiting.



The supplied implementation contains only the basic project structure, and requires completing multiple interdependent components including configuration parsing, token-bucket rate limiting, retry scheduling, circuit-breaker state management, metrics collection, structured logging, and portions of the automated test suite before the application functions correctly.

Challenge Dimension:

Cross-source reasoning by requiring the solver to reconcile behavior described across the configuration specification, architecture/design note, incomplete codebase, and timing-based automated tests.

Multi-item state tracking by coordinating concurrent request execution while simultaneously tracking per-host concurrency, retry attempts, token availability, circuit-breaker state, and request completion across multiple hosts.

Deliverable:

The completed submission should include:

Source implementation of the configurable asynchronous rate limiter.

Configuration file(s) implementing the supplied specification.

Automated unit and integration test suite.

Structured logging output from a representative execution.

Metrics output (JSON or console report).

A short implementation summary (3-5 lines) describing the architecture and key design decisions.

Rubric

Weight Evaluation Criteria

0.05 Project builds and runs using the provided setup instructions

0.15 Configuration is correctly parsed and applied to all hosts

0.15 Per-host concurrency limits are enforced and verified by automated tests

0.15 Token-bucket rate limiting correctly regulates request throughput

0.15 Retry logic implements exponential backoff with jitter according to specification

0.10 Circuit breaker correctly handles failure thresholds and recovery states

0.10 Metrics and structured logs accurately capture request lifecycle and runtime statistics

0.10 Automated tests validate concurrency, timing behavior, retries, and circuit-breaker functionality

0.05 All supplied and newly added automated tests pass without modifying the provided test assertions except where explicitly instructed by the task

📌 Software Engineer - I (India)
🏢 Codefeast
📍 India