pro
Sachin Rajan
AI Engineer building agents, automation & ML systems
New to Contra
Sachin is ready for their next project!
Why I built GradeWise around readiness, not activity
Most interview preparation tools measure activity.
Questions completed.
Hours studied.
Mocks taken.
Topics checked off.
But none of those answer the question that actually matters:
Are you ready for the interview you have?
That became the core idea behind GradeWise.
Instead of treating preparation as a static checklist, I designed GradeWise around an evolving readiness model.
A candidate starts with a specific target: role, company, level, timeline, and curriculum.
From there, the system builds evidence through diagnostics, assessments, practice, review, and other preparation signals.
That creates a loop:
Target → Diagnostic → Evidence → Mastery → Recommendation → New Evidence
The interesting engineering problem is that this cannot simply be an LLM deciding what someone should study next.
The underlying state has to remain reliable as evidence changes.
So the AI layer sits inside a broader system of structured data, deterministic rules, validation, historical state, and explicit workflows.
That distinction matters.
A prototype can generate a recommendation.
A product has to explain where that recommendation came from, update it when new evidence appears, and stay consistent when multiple events affect the same learner state.
The goal with GradeWise is simple:
Prepare for the interview you actually have.
Not more content.
Not more checkboxes.
A clearer picture of where you stand, what you are weak at, and what you should do next.
0
14
Gradewise - AI-powered interview mastery platform
0
0
Built Forge Crew, a local-first multi-agent software engineering workspace for working across large existing codebases.
A central orchestrator coordinates specialist engineering agents for architecture, backend, frontend, databases, deployment, testing and review. The system maintains durable task context, routes work between agents, executes controlled tools and validations, handles failures and retries, and keeps the user in control through approval gates and reviewable changes.
The goal is to make agentic software development structured, observable and auditable rather than allowing autonomous coding agents to modify repositories as a black box.
0
19
Built Aegis, an autonomous codebase hygiene and software-engineering agent that connects to GitHub, scans repositories using deterministic analyzers, classifies issues by risk, and uses an LLM-powered remediation workflow to propose and validate fixes.
The system combines deterministic scanning with AI reasoning rather than relying on an LLM alone. It supports diff-aware analysis, asynchronous workers, automated test execution, GitHub integration, and pull-request based remediation.
0
24