Alan Binu

Junior AI Engineer building humanoid robots (NVIDIA GR00T, ROS2), real-time CV & multi-agent LLM systems. Sharing what I build + break.

[Alan Binu @AlanBin44785396]

Kaspersky tested 1 LLM call vs a 4-agent chain for auto-generating security alert filters.
Accuracy was almost identical, so they shipped the single call: 99% fewer false positives.
When has multi-agent actually beaten one good prompt for you?

[Alan Binu @AlanBin44785396]

@EmmerichMarcelo Agree, detection after the run just starts the clock. Agent-run CI makes it worse since agents happily echo env vars while debugging. Short-lived OIDC tokens instead of stored secrets limit what a leak can do.

[Alan Binu @AlanBin44785396]

@Alacritic_Super @andresvarela Makes sense, the Mac has way more headroom for vision. Then wifi becomes the weak link. Does tracking hold up when the network gets busy, or does it start to lag?

[Alan Binu @AlanBin44785396]

@andresvarela @Alacritic_Super Thanks, going through the face tracking one first. Curious how much delay the phone-to-Mac hop adds before the head actually turns.

[Alan Binu @AlanBin44785396]

@Alacritic_Super Impedance control on an open-source arm is the part that matters for contact tasks like wiping or insertion. Usual catch is gearbox friction making torque estimates noisy. Does it use joint torque sensors or motor current?

[Alan Binu @AlanBin44785396]

@SCR01111 Most 'the LLM is dumb' bugs in RAG agents turn out to be retrieval: the right chunk never made it into context. Logging the retrieved chunks next to each answer kills half the guesswork. Do you trace retrieval scores too?

[Alan Binu @AlanBin44785396]

@mecid Still read them, maybe more now. The agent writes code fast, but you need to know what good looks like to catch it using a deprecated API or a pattern that was fine three years ago.

[Alan Binu @AlanBin44785396]

@techniahqrobot Magnetic feet plus welding is a tricky combo. Hull plates are curved, painted and often rusty, and all of that cuts magnetic grip. Curious if it re-checks foot adhesion before every step or only before a weld.

[Alan Binu @AlanBin44785396]

Source: Compass 1 by Archon Robotics with HKU, Shanghai Innovation Institute, Spatial Robotics and Huawei. Numbers from the project page (ablation on Office + Street, 160 trials each): https://t.co/n2zmgxGMPX

[Alan Binu @AlanBin44785396]

Compass 1 navigates with ~2 min of visual memory from a single RGB camera: 79.4% success over 320 real trials vs 17.8% for the best baseline.
Wildest ablation: drop video generation and it falls from 78.8% to 46.9%.
Is imagining the future the real trick?

[Alan Binu @AlanBin44785396]

@icanvardar Feels like the opposite to me. Types matter more now, they're the cheapest check that catches an agent's wrong guess about a function before anything runs.

[Alan Binu @AlanBin44785396]

@techniahqrobot The NTT one for me too. 73.1% to 92.3% is a big jump, but it depends where the misses went. If physics predictions mostly fix grasp failures, it should transfer. If it's tuned to one setup, less so.

[Alan Binu @AlanBin44785396]

@Josephyala Agents make this worse because they love to 'helpfully' reformat files they pass through. A simple check that flags diffs touching files outside the task's scope would catch most of it before a human even looks.

[Alan Binu @AlanBin44785396]

@rollbar Grouping by code_version with distinct users is a clean check. One gotcha: a rare error can show zero on the new version just because nobody hit that path yet. Do you wait for a minimum number of sessions before calling it fixed?

[Alan Binu @AlanBin44785396]

@andresvarela @Alacritic_Super That split makes a lot more sense. Pico W on the servos keeps motor timing off the phone, and pushing TTS and API calls to the MacBook keeps the phone cool. Is face tracking still on the phone, or does it stream frames to the Mac?

[Alan Binu @AlanBin44785396]

@Alacritic_Super @nio_robotics Using a leg as a tool arm is clever for inspection work. The hard part with shape-changing robots is control: every configuration is basically a different robot to plan for. Curious if they run one policy across shapes or switch controllers per mode.

[Alan Binu @AlanBin44785396]

@EmmerichMarcelo Review capacity is the one that bites first. Five people in a pilot can eyeball every agent PR, at a few hundred the queue grows faster than the reviewers. Did they say how teams decide which changes still need a human?

[Alan Binu @AlanBin44785396]

Hot take: in real-time object counting, the detector is rarely what breaks your numbers. The tracker is.
One ID switch behind an occlusion and the same person gets counted twice.
Do you track IDF1 or HOTA, or just mAP? #ComputerVision

[Alan Binu @AlanBin44785396]

@DivyanshT91162 Real problem. Ask an agent for a modal and you get 80 lines of JS plus a focus trap it probably got wrong, when the native dialog element covers most of it. Agents copy the average of old code, not current best practice.

[Alan Binu @AlanBin44785396]

@codingXdynasty The modern version is an agent saying 'fixed!' while the dev server is still serving the old build. Same problem, just faster. Hard refresh is still step one.

[Alan Binu @AlanBin44785396]

@CodingDevTools The tried/rejected list is the part I'd make mandatory. Status paragraphs read nicely but the next agent still walks into the same wall. Do you have the agent write it, or fill it in yourself at handoff?