DETROIT — For about two years, General Motors engineer Brian Stouffer tried to figure out why faulty ignition switches now linked to at least 13 deaths were causing cars to stall. His quest was thwarted by uncooperative colleagues, inaccurate data and a rotating cast of managers.

Newly released company records depict a low-level engineer struggling to find answers at an automaker in the throes of rebuilding after a 2009 bankruptcy. His bosses on the assignment changed three times in about a year, and he found himself stymied because another engineer had ordered a fix years earlier without the usual paperwork – helping explain why GM took more than a decade to come clean about the defect and recall 2.59 million vehicles, including the Chevrolet Cobalt.

Even as Stouffer chased leads, managers who still play key roles at GM were receiving blunt warnings from safety regulators that they were mishandling recall investigations. One official’s email in 2013 blasting GM for being “too slow to communicate” and “too slow to act” was quickly circulated to product executives and top deputies of Mary Barra, who at the time was GM’s chief product officer and now runs the largest U.S. automaker.

“Typical General Motors,” Maryann Keller, a veteran auto analyst who has written two books on GM, said Sunday. “Where one hand doesn’t know what the other hand is doing, where matters of importance that may cost the company money are sidetracked or dismissed or buried inside the company and people who are supposed to be doing this aren’t given the resources.”

Emails chronicling Stouffer’s journey, which started in 2011, tell just one of the tales in documents released last week by a U.S. House committee investigating the recall. The dead-ends he encountered underscore the confusion caused by a different engineer’s decision in 2006 to change the faulty switch without giving it a new part number.

GM has said it first detected the ignition flaw in 2001 in a Saturn Ion and considered it fixed. The issue re-emerged in 2003 in Saturn models and in 2005 model-year Cobalts. The National Highway Traffic Safety Administration first asked GM in 2007 about problems with the switch that was finally recalled eight weeks ago, according to the documents.

In August 2011, Stouffer, an assessment engineer in GM’s product investigation department, took over the review of the ignition issue that had already been underway for a year. He didn’t realize at the time that GM engineer Ray DeGiorgio, whose team designed the switch, had already quietly greenlighted a fix through supplier Delphi Automotive Plc that began showing up in some 2007-model Cobalts and other small cars.

In June 2012, Stouffer received an email with a report of all of the changes made to the Cobalt steering columns through the 2008 model year, and it didn’t describe changes that would have affected the ignition-lock housing on the steering column or the amount of torque needed to turn the key.

By Sept. 6, 2012, he was seeking additional support and an escalation of the probe from a special investigating team called Red X: “I am trying to understand why 2008 customers no longer report vehicle stalling,” he wrote in an email.

The document trail shows Stouffer was perceiving resistance, or at least a lack of helpfulness, to the probe from the division where DeGiorgio had led the design of the ignition switch. “Early in the discussion we got pushback from Electrical regarding their need to be involved,” Stouffer wrote in an email on Dec. 13, 2013, to a GM investigator asking him for details.

Stouffer’s colleagues “didn’t treat him with a lot of respect,” Lance Cooper, a Marietta, Ga.-based lawyer who represented the family of a crash victim, said in an interview last week. In May 2013, Stouffer – then in his 29th year at GM – gave a deposition in Cooper’s wrongful death lawsuit, the case that uncovered that a faulty ignition switch had been changed.

Stouffer retired in February, according to his LinkedIn profile. He didn’t return calls left on a phone number associated with his name and GM hasn’t made him available for comment.

Kevin Kelly, a GM spokesman, didn’t respond Sunday to questions about the documents. Barra, who became chief executive officer in January, said she learned of an analysis about the stalling cars in December and the recall decision on Jan. 31. She has said GM failed to act quickly enough and apologized for the deaths. Last week she suspended DeGiorgio and another engineer, Gary Altman.

Stouffer, who earned a mechanical engineering degree from Iowa State University and an MBA from Michigan’s Oakland University, was working against two forces: a limited number of incidents, making it harder to draw conclusions and sound an alarm, and a revolving cast of executives assigned to the probe – three in about one year.

In the deposition for Cooper, Stouffer said he looked through GM’s database of 500,000 vehicles and was able to find about 100 complaints that seemed to meet the criteria. “I have 100 complaints for that, which is a very, very low complaint rate,” he said in the deposition.

Cooper, the attorney, didn’t let Stouffer off the hook during the deposition, suggesting he could have pressed the investigation more forcefully. He confronted Stouffer with the idea that the switch may have been changed in later models and asked why he never compared the old switches with existing ones found at dealerships.